An improved advertising CTR prediction approach based on the fuzzy deep neural network.

PubMed

Jiang, Zilong; Gao, Shu; Li, Mingjiang

2018-01-01

Combining a deep neural network with fuzzy theory, this paper proposes an advertising click-through rate (CTR) prediction approach based on a fuzzy deep neural network (FDNN). In this approach, fuzzy Gaussian-Bernoulli restricted Boltzmann machine (FGBRBM) is first applied to input raw data from advertising datasets. Next, fuzzy restricted Boltzmann machine (FRBM) is used to construct the fuzzy deep belief network (FDBN) with the unsupervised method layer by layer. Finally, fuzzy logistic regression (FLR) is utilized for modeling the CTR. The experimental results show that the proposed FDNN model outperforms several baseline models in terms of both data representation capability and robustness in advertising click log datasets with noise.

Robust hepatic vessel segmentation using multi deep convolution network

NASA Astrophysics Data System (ADS)

Kitrungrotsakul, Titinunt; Han, Xian-Hua; Iwamoto, Yutaro; Foruzan, Amir Hossein; Lin, Lanfen; Chen, Yen-Wei

2017-03-01

Extraction of blood vessels of the organ is a challenging task in the area of medical image processing. It is really difficult to get accurate vessel segmentation results even with manually labeling by human being. The difficulty of vessels segmentation is the complicated structure of blood vessels and its large variations that make them hard to recognize. In this paper, we present deep artificial neural network architecture to automatically segment the hepatic vessels from computed tomography (CT) image. We proposed novel deep neural network (DNN) architecture for vessel segmentation from a medical CT volume, which consists of three deep convolution neural networks to extract features from difference planes of CT data. The three networks have share features at the first convolution layer but will separately learn their own features in the second layer. All three networks will join again at the top layer. To validate effectiveness and efficiency of our proposed method, we conduct experiments on 12 CT volumes which training data are randomly generate from 5 CT volumes and 7 using for test. Our network can yield an average dice coefficient 0.830, while 3D deep convolution neural network can yield around 0.7 and multi-scale can yield only 0.6.

Intrusion Detection System Using Deep Neural Network for In-Vehicle Network Security.

PubMed

Kang, Min-Joo; Kang, Je-Won

2016-01-01

A novel intrusion detection system (IDS) using a deep neural network (DNN) is proposed to enhance the security of in-vehicular network. The parameters building the DNN structure are trained with probability-based feature vectors that are extracted from the in-vehicular network packets. For a given packet, the DNN provides the probability of each class discriminating normal and attack packets, and, thus the sensor can identify any malicious attack to the vehicle. As compared to the traditional artificial neural network applied to the IDS, the proposed technique adopts recent advances in deep learning studies such as initializing the parameters through the unsupervised pre-training of deep belief networks (DBN), therefore improving the detection accuracy. It is demonstrated with experimental results that the proposed technique can provide a real-time response to the attack with a significantly improved detection ratio in controller area network (CAN) bus.

Intrusion Detection System Using Deep Neural Network for In-Vehicle Network Security

PubMed Central

Kang, Min-Joo

2016-01-01

A novel intrusion detection system (IDS) using a deep neural network (DNN) is proposed to enhance the security of in-vehicular network. The parameters building the DNN structure are trained with probability-based feature vectors that are extracted from the in-vehicular network packets. For a given packet, the DNN provides the probability of each class discriminating normal and attack packets, and, thus the sensor can identify any malicious attack to the vehicle. As compared to the traditional artificial neural network applied to the IDS, the proposed technique adopts recent advances in deep learning studies such as initializing the parameters through the unsupervised pre-training of deep belief networks (DBN), therefore improving the detection accuracy. It is demonstrated with experimental results that the proposed technique can provide a real-time response to the attack with a significantly improved detection ratio in controller area network (CAN) bus. PMID:27271802

Deep neural network for traffic sign recognition systems: An analysis of spatial transformers and stochastic optimisation methods.

PubMed

Arcos-García, Álvaro; Álvarez-García, Juan A; Soria-Morillo, Luis M

2018-03-01

This paper presents a Deep Learning approach for traffic sign recognition systems. Several classification experiments are conducted over publicly available traffic sign datasets from Germany and Belgium using a Deep Neural Network which comprises Convolutional layers and Spatial Transformer Networks. Such trials are built to measure the impact of diverse factors with the end goal of designing a Convolutional Neural Network that can improve the state-of-the-art of traffic sign classification task. First, different adaptive and non-adaptive stochastic gradient descent optimisation algorithms such as SGD, SGD-Nesterov, RMSprop and Adam are evaluated. Subsequently, multiple combinations of Spatial Transformer Networks placed at distinct positions within the main neural network are analysed. The recognition rate of the proposed Convolutional Neural Network reports an accuracy of 99.71% in the German Traffic Sign Recognition Benchmark, outperforming previous state-of-the-art methods and also being more efficient in terms of memory requirements. Copyright © 2018 Elsevier Ltd. All rights reserved.

How Deep Neural Networks Can Improve Emotion Recognition on Video Data

DTIC Science & Technology

2016-09-25

HOW DEEP NEURAL NETWORKS CAN IMPROVE EMOTION RECOGNITION ON VIDEO DATA Pooya Khorrami1 , Tom Le Paine1, Kevin Brady2, Charlie Dagli2, Thomas S...this work, we present a system that per- forms emotion recognition on video data using both con- volutional neural networks (CNNs) and recurrent...neural net- works (RNNs). We present our findings on videos from the Audio/Visual+Emotion Challenge (AV+EC2015). In our experiments, we analyze the effects

Classifying the molecular functions of Rab GTPases in membrane trafficking using deep convolutional neural networks.

PubMed

Le, Nguyen-Quoc-Khanh; Ho, Quang-Thai; Ou, Yu-Yen

2018-06-13

Deep learning has been increasingly used to solve a number of problems with state-of-the-art performance in a wide variety of fields. In biology, deep learning can be applied to reduce feature extraction time and achieve high levels of performance. In our present work, we apply deep learning via two-dimensional convolutional neural networks and position-specific scoring matrices to classify Rab protein molecules, which are main regulators in membrane trafficking for transferring proteins and other macromolecules throughout the cell. The functional loss of specific Rab molecular functions has been implicated in a variety of human diseases, e.g., choroideremia, intellectual disabilities, cancer. Therefore, creating a precise model for classifying Rabs is crucial in helping biologists understand the molecular functions of Rabs and design drug targets according to such specific human disease information. We constructed a robust deep neural network for classifying Rabs that achieved an accuracy of 99%, 99.5%, 96.3%, and 97.6% for each of four specific molecular functions. Our approach demonstrates superior performance to traditional artificial neural networks. Therefore, from our proposed study, we provide both an effective tool for classifying Rab proteins and a basis for further research that can improve the performance of biological modeling using deep neural networks. Copyright © 2018 Elsevier Inc. All rights reserved.

Sentiment analysis: a comparison of deep learning neural network algorithm with SVM and naϊve Bayes for Indonesian text

NASA Astrophysics Data System (ADS)

Calvin Frans Mariel, Wahyu; Mariyah, Siti; Pramana, Setia

2018-03-01

Deep learning is a new era of machine learning techniques that essentially imitate the structure and function of the human brain. It is a development of deeper Artificial Neural Network (ANN) that uses more than one hidden layer. Deep Learning Neural Network has a great ability on recognizing patterns from various data types such as picture, audio, text, and many more. In this paper, the authors tries to measure that algorithm’s ability by applying it into the text classification. The classification task herein is done by considering the content of sentiment in a text which is also called as sentiment analysis. By using several combinations of text preprocessing and feature extraction techniques, we aim to compare the precise modelling results of Deep Learning Neural Network with the other two commonly used algorithms, the Naϊve Bayes and Support Vector Machine (SVM). This algorithm comparison uses Indonesian text data with balanced and unbalanced sentiment composition. Based on the experimental simulation, Deep Learning Neural Network clearly outperforms the Naϊve Bayes and SVM and offers a better F-1 Score while for the best feature extraction technique which improves that modelling result is Bigram.

Training Deep Spiking Neural Networks Using Backpropagation.

PubMed

Lee, Jun Haeng; Delbruck, Tobi; Pfeiffer, Michael

2016-01-01

Deep spiking neural networks (SNNs) hold the potential for improving the latency and energy efficiency of deep neural networks through data-driven event-based computation. However, training such networks is difficult due to the non-differentiable nature of spike events. In this paper, we introduce a novel technique, which treats the membrane potentials of spiking neurons as differentiable signals, where discontinuities at spike times are considered as noise. This enables an error backpropagation mechanism for deep SNNs that follows the same principles as in conventional deep networks, but works directly on spike signals and membrane potentials. Compared with previous methods relying on indirect training and conversion, our technique has the potential to capture the statistics of spikes more precisely. We evaluate the proposed framework on artificially generated events from the original MNIST handwritten digit benchmark, and also on the N-MNIST benchmark recorded with an event-based dynamic vision sensor, in which the proposed method reduces the error rate by a factor of more than three compared to the best previous SNN, and also achieves a higher accuracy than a conventional convolutional neural network (CNN) trained and tested on the same data. We demonstrate in the context of the MNIST task that thanks to their event-driven operation, deep SNNs (both fully connected and convolutional) trained with our method achieve accuracy equivalent with conventional neural networks. In the N-MNIST example, equivalent accuracy is achieved with about five times fewer computational operations.

A Composite Model of Wound Segmentation Based on Traditional Methods and Deep Neural Networks

PubMed Central

Wang, Changjian; Liu, Xiaohui; Jin, Shiyao

2018-01-01

Wound segmentation plays an important supporting role in the wound observation and wound healing. Current methods of image segmentation include those based on traditional process of image and those based on deep neural networks. The traditional methods use the artificial image features to complete the task without large amounts of labeled data. Meanwhile, the methods based on deep neural networks can extract the image features effectively without the artificial design, but lots of training data are required. Combined with the advantages of them, this paper presents a composite model of wound segmentation. The model uses the skin with wound detection algorithm we designed in the paper to highlight image features. Then, the preprocessed images are segmented by deep neural networks. And semantic corrections are applied to the segmentation results at last. The model shows a good performance in our experiment. PMID:29955227

Pattern learning with deep neural networks in EMG-based speech recognition.

PubMed

Wand, Michael; Schultz, Tanja

2014-01-01

We report on classification of phones and phonetic features from facial electromyographic (EMG) data, within the context of our EMG-based Silent Speech interface. In this paper we show that a Deep Neural Network can be used to perform this classification task, yielding a significant improvement over conventional Gaussian Mixture models. Our central contribution is the visualization of patterns which are learned by the neural network. With increasing network depth, these patterns represent more and more intricate electromyographic activity.

Deep Constrained Siamese Hash Coding Network and Load-Balanced Locality-Sensitive Hashing for Near Duplicate Image Detection.

PubMed

Hu, Weiming; Fan, Yabo; Xing, Junliang; Sun, Liang; Cai, Zhaoquan; Maybank, Stephen

2018-09-01

We construct a new efficient near duplicate image detection method using a hierarchical hash code learning neural network and load-balanced locality-sensitive hashing (LSH) indexing. We propose a deep constrained siamese hash coding neural network combined with deep feature learning. Our neural network is able to extract effective features for near duplicate image detection. The extracted features are used to construct a LSH-based index. We propose a load-balanced LSH method to produce load-balanced buckets in the hashing process. The load-balanced LSH significantly reduces the query time. Based on the proposed load-balanced LSH, we design an effective and feasible algorithm for near duplicate image detection. Extensive experiments on three benchmark data sets demonstrate the effectiveness of our deep siamese hash encoding network and load-balanced LSH.

An analysis of image storage systems for scalable training of deep neural networks

DOE Office of Scientific and Technical Information (OSTI.GOV)

Lim, Seung-Hwan; Young, Steven R; Patton, Robert M

This study presents a principled empirical evaluation of image storage systems for training deep neural networks. We employ the Caffe deep learning framework to train neural network models for three different data sets, MNIST, CIFAR-10, and ImageNet. While training the models, we evaluate five different options to retrieve training image data: (1) PNG-formatted image files on local file system; (2) pushing pixel arrays from image files into a single HDF5 file on local file system; (3) in-memory arrays to hold the pixel arrays in Python and C++; (4) loading the training data into LevelDB, a log-structured merge tree based key-valuemore » storage; and (5) loading the training data into LMDB, a B+tree based key-value storage. The experimental results quantitatively highlight the disadvantage of using normal image files on local file systems to train deep neural networks and demonstrate reliable performance with key-value storage based storage systems. When training a model on the ImageNet dataset, the image file option was more than 17 times slower than the key-value storage option. Along with measurements on training time, this study provides in-depth analysis on the cause of performance advantages/disadvantages of each back-end to train deep neural networks. We envision the provided measurements and analysis will shed light on the optimal way to architect systems for training neural networks in a scalable manner.« less

Simple techniques for improving deep neural network outcomes on commodity hardware

NASA Astrophysics Data System (ADS)

Colina, Nicholas Christopher A.; Perez, Carlos E.; Paraan, Francis N. C.

2017-08-01

We benchmark improvements in the performance of deep neural networks (DNN) on the MNIST data test upon imple-menting two simple modifications to the algorithm that have little overhead computational cost. First is GPU parallelization on a commodity graphics card, and second is initializing the DNN with random orthogonal weight matrices prior to optimization. Eigenspectra analysis of the weight matrices reveal that the initially orthogonal matrices remain nearly orthogonal after training. The probability distributions from which these orthogonal matrices are drawn are also shown to significantly affect the performance of these deep neural networks.

Deep Neural Network for Structural Prediction and Lane Detection in Traffic Scene.

PubMed

Li, Jun; Mei, Xue; Prokhorov, Danil; Tao, Dacheng

2017-03-01

Hierarchical neural networks have been shown to be effective in learning representative image features and recognizing object classes. However, most existing networks combine the low/middle level cues for classification without accounting for any spatial structures. For applications such as understanding a scene, how the visual cues are spatially distributed in an image becomes essential for successful analysis. This paper extends the framework of deep neural networks by accounting for the structural cues in the visual signals. In particular, two kinds of neural networks have been proposed. First, we develop a multitask deep convolutional network, which simultaneously detects the presence of the target and the geometric attributes (location and orientation) of the target with respect to the region of interest. Second, a recurrent neuron layer is adopted for structured visual detection. The recurrent neurons can deal with the spatial distribution of visible cues belonging to an object whose shape or structure is difficult to explicitly define. Both the networks are demonstrated by the practical task of detecting lane boundaries in traffic scenes. The multitask convolutional neural network provides auxiliary geometric information to help the subsequent modeling of the given lane structures. The recurrent neural network automatically detects lane boundaries, including those areas containing no marks, without any explicit prior knowledge or secondary modeling.

A Hybrid Spectral Clustering and Deep Neural Network Ensemble Algorithm for Intrusion Detection in Sensor Networks

PubMed Central

Ma, Tao; Wang, Fen; Cheng, Jianjun; Yu, Yang; Chen, Xiaoyun

2016-01-01

The development of intrusion detection systems (IDS) that are adapted to allow routers and network defence systems to detect malicious network traffic disguised as network protocols or normal access is a critical challenge. This paper proposes a novel approach called SCDNN, which combines spectral clustering (SC) and deep neural network (DNN) algorithms. First, the dataset is divided into k subsets based on sample similarity using cluster centres, as in SC. Next, the distance between data points in a testing set and the training set is measured based on similarity features and is fed into the deep neural network algorithm for intrusion detection. Six KDD-Cup99 and NSL-KDD datasets and a sensor network dataset were employed to test the performance of the model. These experimental results indicate that the SCDNN classifier not only performs better than backpropagation neural network (BPNN), support vector machine (SVM), random forest (RF) and Bayes tree models in detection accuracy and the types of abnormal attacks found. It also provides an effective tool of study and analysis of intrusion detection in large networks. PMID:27754380

A Hybrid Spectral Clustering and Deep Neural Network Ensemble Algorithm for Intrusion Detection in Sensor Networks.

PubMed

Ma, Tao; Wang, Fen; Cheng, Jianjun; Yu, Yang; Chen, Xiaoyun

2016-10-13

The development of intrusion detection systems (IDS) that are adapted to allow routers and network defence systems to detect malicious network traffic disguised as network protocols or normal access is a critical challenge. This paper proposes a novel approach called SCDNN, which combines spectral clustering (SC) and deep neural network (DNN) algorithms. First, the dataset is divided into k subsets based on sample similarity using cluster centres, as in SC. Next, the distance between data points in a testing set and the training set is measured based on similarity features and is fed into the deep neural network algorithm for intrusion detection. Six KDD-Cup99 and NSL-KDD datasets and a sensor network dataset were employed to test the performance of the model. These experimental results indicate that the SCDNN classifier not only performs better than backpropagation neural network (BPNN), support vector machine (SVM), random forest (RF) and Bayes tree models in detection accuracy and the types of abnormal attacks found. It also provides an effective tool of study and analysis of intrusion detection in large networks.

Reynolds averaged turbulence modelling using deep neural networks with embedded invariance

DOE PAGES

Ling, Julia; Kurzawski, Andrew; Templeton, Jeremy

2016-10-18

There exists significant demand for improved Reynolds-averaged Navier–Stokes (RANS) turbulence models that are informed by and can represent a richer set of turbulence physics. This paper presents a method of using deep neural networks to learn a model for the Reynolds stress anisotropy tensor from high-fidelity simulation data. A novel neural network architecture is proposed which uses a multiplicative layer with an invariant tensor basis to embed Galilean invariance into the predicted anisotropy tensor. It is demonstrated that this neural network architecture provides improved prediction accuracy compared with a generic neural network architecture that does not embed this invariance property.more » Furthermore, the Reynolds stress anisotropy predictions of this invariant neural network are propagated through to the velocity field for two test cases. For both test cases, significant improvement versus baseline RANS linear eddy viscosity and nonlinear eddy viscosity models is demonstrated.« less

Reynolds averaged turbulence modelling using deep neural networks with embedded invariance

DOE Office of Scientific and Technical Information (OSTI.GOV)

Ling, Julia; Kurzawski, Andrew; Templeton, Jeremy

There exists significant demand for improved Reynolds-averaged Navier–Stokes (RANS) turbulence models that are informed by and can represent a richer set of turbulence physics. This paper presents a method of using deep neural networks to learn a model for the Reynolds stress anisotropy tensor from high-fidelity simulation data. A novel neural network architecture is proposed which uses a multiplicative layer with an invariant tensor basis to embed Galilean invariance into the predicted anisotropy tensor. It is demonstrated that this neural network architecture provides improved prediction accuracy compared with a generic neural network architecture that does not embed this invariance property.more » Furthermore, the Reynolds stress anisotropy predictions of this invariant neural network are propagated through to the velocity field for two test cases. For both test cases, significant improvement versus baseline RANS linear eddy viscosity and nonlinear eddy viscosity models is demonstrated.« less

On the complexity of neural network classifiers: a comparison between shallow and deep architectures.

PubMed

Bianchini, Monica; Scarselli, Franco

2014-08-01

Recently, researchers in the artificial neural network field have focused their attention on connectionist models composed by several hidden layers. In fact, experimental results and heuristic considerations suggest that deep architectures are more suitable than shallow ones for modern applications, facing very complex problems, e.g., vision and human language understanding. However, the actual theoretical results supporting such a claim are still few and incomplete. In this paper, we propose a new approach to study how the depth of feedforward neural networks impacts on their ability in implementing high complexity functions. First, a new measure based on topological concepts is introduced, aimed at evaluating the complexity of the function implemented by a neural network, used for classification purposes. Then, deep and shallow neural architectures with common sigmoidal activation functions are compared, by deriving upper and lower bounds on their complexity, and studying how the complexity depends on the number of hidden units and the used activation function. The obtained results seem to support the idea that deep networks actually implements functions of higher complexity, so that they are able, with the same number of resources, to address more difficult problems.

Deep learning for brain tumor classification

NASA Astrophysics Data System (ADS)

Paul, Justin S.; Plassard, Andrew J.; Landman, Bennett A.; Fabbri, Daniel

2017-03-01

Recent research has shown that deep learning methods have performed well on supervised machine learning, image classification tasks. The purpose of this study is to apply deep learning methods to classify brain images with different tumor types: meningioma, glioma, and pituitary. A dataset was publicly released containing 3,064 T1-weighted contrast enhanced MRI (CE-MRI) brain images from 233 patients with either meningioma, glioma, or pituitary tumors split across axial, coronal, or sagittal planes. This research focuses on the 989 axial images from 191 patients in order to avoid confusing the neural networks with three different planes containing the same diagnosis. Two types of neural networks were used in classification: fully connected and convolutional neural networks. Within these two categories, further tests were computed via the augmentation of the original 512×512 axial images. Training neural networks over the axial data has proven to be accurate in its classifications with an average five-fold cross validation of 91.43% on the best trained neural network. This result demonstrates that a more general method (i.e. deep learning) can outperform specialized methods that require image dilation and ring-forming subregions on tumors.

Deep convolutional neural networks for automatic classification of gastric carcinoma using whole slide images in digital histopathology.

PubMed

Sharma, Harshita; Zerbe, Norman; Klempert, Iris; Hellwich, Olaf; Hufnagl, Peter

2017-11-01

Deep learning using convolutional neural networks is an actively emerging field in histological image analysis. This study explores deep learning methods for computer-aided classification in H&E stained histopathological whole slide images of gastric carcinoma. An introductory convolutional neural network architecture is proposed for two computerized applications, namely, cancer classification based on immunohistochemical response and necrosis detection based on the existence of tumor necrosis in the tissue. Classification performance of the developed deep learning approach is quantitatively compared with traditional image analysis methods in digital histopathology requiring prior computation of handcrafted features, such as statistical measures using gray level co-occurrence matrix, Gabor filter-bank responses, LBP histograms, gray histograms, HSV histograms and RGB histograms, followed by random forest machine learning. Additionally, the widely known AlexNet deep convolutional framework is comparatively analyzed for the corresponding classification problems. The proposed convolutional neural network architecture reports favorable results, with an overall classification accuracy of 0.6990 for cancer classification and 0.8144 for necrosis detection. Copyright © 2017 Elsevier Ltd. All rights reserved.

Deep convolutional neural networks for estimating porous material parameters with ultrasound tomography

NASA Astrophysics Data System (ADS)

Lähivaara, Timo; Kärkkäinen, Leo; Huttunen, Janne M. J.; Hesthaven, Jan S.

2018-02-01

We study the feasibility of data based machine learning applied to ultrasound tomography to estimate water-saturated porous material parameters. In this work, the data to train the neural networks is simulated by solving wave propagation in coupled poroviscoelastic-viscoelastic-acoustic media. As the forward model, we consider a high-order discontinuous Galerkin method while deep convolutional neural networks are used to solve the parameter estimation problem. In the numerical experiment, we estimate the material porosity and tortuosity while the remaining parameters which are of less interest are successfully marginalized in the neural networks-based inversion. Computational examples confirms the feasibility and accuracy of this approach.

Using Deep Learning for Gamma Ray Source Detection at the First G-APD Cherenkov Telescope (FACT)

NASA Astrophysics Data System (ADS)

Bieker, Jacob

2018-06-01

Finding gamma-ray sources is of paramount importance for Imaging Air Cherenkov Telescopes (IACT). This study looks at using deep neural networks on data from the First G-APD Cherenkov Telescope (FACT) as a proof-of-concept of finding gamma-ray sources with deep learning for the upcoming Cherenkov Telescope Array (CTA). In this study, FACT’s individual photon level observation data from the last 5 years was used with convolutional neural networks to determine if one or more sources were present. The neural networks used various architectures to determine which architectures were most successful in finding sources. Neural networks offer a promising method for finding faint and extended gamma-ray sources for IACTs. With further improvement and modifications, they offer a compelling method for source detection for the next generation of IACTs.

Maximum entropy methods for extracting the learned features of deep neural networks.

PubMed

Finnegan, Alex; Song, Jun S

2017-10-01

New architectures of multilayer artificial neural networks and new methods for training them are rapidly revolutionizing the application of machine learning in diverse fields, including business, social science, physical sciences, and biology. Interpreting deep neural networks, however, currently remains elusive, and a critical challenge lies in understanding which meaningful features a network is actually learning. We present a general method for interpreting deep neural networks and extracting network-learned features from input data. We describe our algorithm in the context of biological sequence analysis. Our approach, based on ideas from statistical physics, samples from the maximum entropy distribution over possible sequences, anchored at an input sequence and subject to constraints implied by the empirical function learned by a network. Using our framework, we demonstrate that local transcription factor binding motifs can be identified from a network trained on ChIP-seq data and that nucleosome positioning signals are indeed learned by a network trained on chemical cleavage nucleosome maps. Imposing a further constraint on the maximum entropy distribution also allows us to probe whether a network is learning global sequence features, such as the high GC content in nucleosome-rich regions. This work thus provides valuable mathematical tools for interpreting and extracting learned features from feed-forward neural networks.

Using deep learning in image hyper spectral segmentation, classification, and detection

NASA Astrophysics Data System (ADS)

Zhao, Xiuying; Su, Zhenyu

2018-02-01

Recent years have shown that deep learning neural networks are a valuable tool in the field of computer vision. Deep learning method can be used in applications like remote sensing such as Land cover Classification, Detection of Vehicle in Satellite Images, Hyper spectral Image classification. This paper addresses the use of the deep learning artificial neural network in Satellite image segmentation. Image segmentation plays an important role in image processing. The hue of the remote sensing image often has a large hue difference, which will result in the poor display of the images in the VR environment. Image segmentation is a pre processing technique applied to the original images and splits the image into many parts which have different hue to unify the color. Several computational models based on supervised, unsupervised, parametric, probabilistic region based image segmentation techniques have been proposed. Recently, one of the machine learning technique known as, deep learning with convolution neural network has been widely used for development of efficient and automatic image segmentation models. In this paper, we focus on study of deep neural convolution network and its variants for automatic image segmentation rather than traditional image segmentation strategies.

Neural network based satellite tracking for deep space applications

NASA Technical Reports Server (NTRS)

Amoozegar, F.; Ruggier, C.

2003-01-01

The objective of this paper is to provide a survey of neural network trends as applied to the tracking of spacecrafts in deep space at Ka-band under various weather conditions and examine the trade-off between tracing accuracy and communication link performance.

SYNAPTIC DEPRESSION IN DEEP NEURAL NETWORKS FOR SPEECH PROCESSING.

PubMed

Zhang, Wenhao; Li, Hanyu; Yang, Minda; Mesgarani, Nima

2016-03-01

A characteristic property of biological neurons is their ability to dynamically change the synaptic efficacy in response to variable input conditions. This mechanism, known as synaptic depression, significantly contributes to the formation of normalized representation of speech features. Synaptic depression also contributes to the robust performance of biological systems. In this paper, we describe how synaptic depression can be modeled and incorporated into deep neural network architectures to improve their generalization ability. We observed that when synaptic depression is added to the hidden layers of a neural network, it reduces the effect of changing background activity in the node activations. In addition, we show that when synaptic depression is included in a deep neural network trained for phoneme classification, the performance of the network improves under noisy conditions not included in the training phase. Our results suggest that more complete neuron models may further reduce the gap between the biological performance and artificial computing, resulting in networks that better generalize to novel signal conditions.

Automatic Classification of volcano-seismic events based on Deep Neural Networks.

NASA Astrophysics Data System (ADS)

Titos Luzón, M.; Bueno Rodriguez, A.; Garcia Martinez, L.; Benitez, C.; Ibáñez, J. M.

2017-12-01

Seismic monitoring of active volcanoes is a popular remote sensing technique to detect seismic activity, often associated to energy exchanges between the volcano and the environment. As a result, seismographs register a wide range of volcano-seismic signals that reflect the nature and underlying physics of volcanic processes. Machine learning and signal processing techniques provide an appropriate framework to analyze such data. In this research, we propose a new classification framework for seismic events based on deep neural networks. Deep neural networks are composed by multiple processing layers, and can discover intrinsic patterns from the data itself. Internal parameters can be initialized using a greedy unsupervised pre-training stage, leading to an efficient training of fully connected architectures. We aim to determine the robustness of these architectures as classifiers of seven different types of seismic events recorded at "Volcán de Fuego" (Colima, Mexico). Two deep neural networks with different pre-training strategies are studied: stacked denoising autoencoder and deep belief networks. Results are compared to existing machine learning algorithms (SVM, Random Forest, Multilayer Perceptron). We used 5 LPC coefficients over three non-overlapping segments as training features in order to characterize temporal evolution, avoid redundancy and encode the signal, regardless of its duration. Experimental results show that deep architectures can classify seismic events with higher accuracy than classical algorithms, attaining up to 92% recognition accuracy. Pre-training initialization helps these models to detect events that occur simultaneously in time (such explosions and rockfalls), increase robustness against noisy inputs, and provide better generalization. These results demonstrate deep neural networks are robust classifiers, and can be deployed in real-environments to monitor the seismicity of restless volcanoes.

A Deep Neural Network Model for Rainfall Estimation UsingPolarimetric WSR-88DP Radar Observations

NASA Astrophysics Data System (ADS)

Tan, H.; Chandra, C. V.; Chen, H.

2016-12-01

Rainfall estimation based on radar measurements has been an important topic for a few decades. Generally, radar rainfall estimation is conducted through parametric algorisms such as reflectivity-rainfall relation (i.e., Z-R relation). On the other hand, neural networks are developed for ground rainfall estimation based on radar measurements. This nonparametric method, which takes into account of both radar observations and rainfall measurements from ground rain gauges, has been demonstrated successfully for rainfall rate estimation. However, the neural network-based rainfall estimation is limited in practice due to the model complexity and structure, data quality, as well as different rainfall microphysics. Recently, the deep learning approach has been introduced in pattern recognition and machine learning areas. Compared to traditional neural networks, the deep learning based methodologies have larger number of hidden layers and more complex structure for data representation. Through a hierarchical learning process, the high level structured information and knowledge can be extracted automatically from low level features of the data. In this paper, we introduce a novel deep neural network model for rainfall estimation based on ground polarimetric radar measurements .The model is designed to capture the complex abstractions of radar measurements at different levels using multiple layers feature identification and extraction. The abstractions at different levels can be used independently or fused with other data resource such as satellite-based rainfall products and/or topographic data to represent the rain characteristics at certain location. In particular, the WSR-88DP radar and rain gauge data collected in Dallas - Fort Worth Metroplex and Florida are used extensively to train the model, and for demonstration purposes. Quantitative evaluation of the deep neural network based rainfall products will also be presented, which is based on an independent rain gauge network.

ChemNet: A Transferable and Generalizable Deep Neural Network for Small-Molecule Property Prediction

DOE Office of Scientific and Technical Information (OSTI.GOV)

Goh, Garrett B.; Siegel, Charles M.; Vishnu, Abhinav

With access to large datasets, deep neural networks through representation learning have been able to identify patterns from raw data, achieving human-level accuracy in image and speech recognition tasks. However, in chemistry, availability of large standardized and labelled datasets is scarce, and with a multitude of chemical properties of interest, chemical data is inherently small and fragmented. In this work, we explore transfer learning techniques in conjunction with the existing Chemception CNN model, to create a transferable and generalizable deep neural network for small-molecule property prediction. Our latest model, ChemNet learns in a semi-supervised manner from inexpensive labels computed frommore » the ChEMBL database. When fine-tuned to the Tox21, HIV and FreeSolv dataset, which are 3 separate chemical tasks that ChemNet was not originally trained on, we demonstrate that ChemNet exceeds the performance of existing Chemception models, contemporary MLP models that trains on molecular fingerprints, and it matches the performance of the ConvGraph algorithm, the current state-of-the-art. Furthermore, as ChemNet has been pre-trained on a large diverse chemical database, it can be used as a universal “plug-and-play” deep neural network, which accelerates the deployment of deep neural networks for the prediction of novel small-molecule chemical properties.« less

Applications of deep convolutional neural networks to digitized natural history collections.

PubMed

Schuettpelz, Eric; Frandsen, Paul B; Dikow, Rebecca B; Brown, Abel; Orli, Sylvia; Peters, Melinda; Metallo, Adam; Funk, Vicki A; Dorr, Laurence J

2017-01-01

Natural history collections contain data that are critical for many scientific endeavors. Recent efforts in mass digitization are generating large datasets from these collections that can provide unprecedented insight. Here, we present examples of how deep convolutional neural networks can be applied in analyses of imaged herbarium specimens. We first demonstrate that a convolutional neural network can detect mercury-stained specimens across a collection with 90% accuracy. We then show that such a network can correctly distinguish two morphologically similar plant families 96% of the time. Discarding the most challenging specimen images increases accuracy to 94% and 99%, respectively. These results highlight the importance of mass digitization and deep learning approaches and reveal how they can together deliver powerful new investigative tools.

Feature to prototype transition in neural networks

NASA Astrophysics Data System (ADS)

Krotov, Dmitry; Hopfield, John

Models of associative memory with higher order (higher than quadratic) interactions, and their relationship to neural networks used in deep learning are discussed. Associative memory is conventionally described by recurrent neural networks with dynamical convergence to stable points. Deep learning typically uses feedforward neural nets without dynamics. However, a simple duality relates these two different views when applied to problems of pattern classification. From the perspective of associative memory such models deserve attention because they make it possible to store a much larger number of memories, compared to the quadratic case. In the dual description, these models correspond to feedforward neural networks with one hidden layer and unusual activation functions transmitting the activities of the visible neurons to the hidden layer. These activation functions are rectified polynomials of a higher degree rather than the rectified linear functions used in deep learning. The network learns representations of the data in terms of features for rectified linear functions, but as the power in the activation function is increased there is a gradual shift to a prototype-based representation, the two extreme regimes of pattern recognition known in cognitive psychology. Simons Center for Systems Biology.

A renaissance of neural networks in drug discovery.

PubMed

Baskin, Igor I; Winkler, David; Tetko, Igor V

2016-08-01

Neural networks are becoming a very popular method for solving machine learning and artificial intelligence problems. The variety of neural network types and their application to drug discovery requires expert knowledge to choose the most appropriate approach. In this review, the authors discuss traditional and newly emerging neural network approaches to drug discovery. Their focus is on backpropagation neural networks and their variants, self-organizing maps and associated methods, and a relatively new technique, deep learning. The most important technical issues are discussed including overfitting and its prevention through regularization, ensemble and multitask modeling, model interpretation, and estimation of applicability domain. Different aspects of using neural networks in drug discovery are considered: building structure-activity models with respect to various targets; predicting drug selectivity, toxicity profiles, ADMET and physicochemical properties; characteristics of drug-delivery systems and virtual screening. Neural networks continue to grow in importance for drug discovery. Recent developments in deep learning suggests further improvements may be gained in the analysis of large chemical data sets. It's anticipated that neural networks will be more widely used in drug discovery in the future, and applied in non-traditional areas such as drug delivery systems, biologically compatible materials, and regenerative medicine.

Stable architectures for deep neural networks

NASA Astrophysics Data System (ADS)

Haber, Eldad; Ruthotto, Lars

2018-01-01

Deep neural networks have become invaluable tools for supervised machine learning, e.g. classification of text or images. While often offering superior results over traditional techniques and successfully expressing complicated patterns in data, deep architectures are known to be challenging to design and train such that they generalize well to new data. Critical issues with deep architectures are numerical instabilities in derivative-based learning algorithms commonly called exploding or vanishing gradients. In this paper, we propose new forward propagation techniques inspired by systems of ordinary differential equations (ODE) that overcome this challenge and lead to well-posed learning problems for arbitrarily deep networks. The backbone of our approach is our interpretation of deep learning as a parameter estimation problem of nonlinear dynamical systems. Given this formulation, we analyze stability and well-posedness of deep learning and use this new understanding to develop new network architectures. We relate the exploding and vanishing gradient phenomenon to the stability of the discrete ODE and present several strategies for stabilizing deep learning for very deep networks. While our new architectures restrict the solution space, several numerical experiments show their competitiveness with state-of-the-art networks.

Clinical Named Entity Recognition Using Deep Learning Models.

PubMed

Wu, Yonghui; Jiang, Min; Xu, Jun; Zhi, Degui; Xu, Hua

2017-01-01

Clinical Named Entity Recognition (NER) is a critical natural language processing (NLP) task to extract important concepts (named entities) from clinical narratives. Researchers have extensively investigated machine learning models for clinical NER. Recently, there have been increasing efforts to apply deep learning models to improve the performance of current clinical NER systems. This study examined two popular deep learning architectures, the Convolutional Neural Network (CNN) and the Recurrent Neural Network (RNN), to extract concepts from clinical texts. We compared the two deep neural network architectures with three baseline Conditional Random Fields (CRFs) models and two state-of-the-art clinical NER systems using the i2b2 2010 clinical concept extraction corpus. The evaluation results showed that the RNN model trained with the word embeddings achieved a new state-of-the- art performance (a strict F1 score of 85.94%) for the defined clinical NER task, outperforming the best-reported system that used both manually defined and unsupervised learning features. This study demonstrates the advantage of using deep neural network architectures for clinical concept extraction, including distributed feature representation, automatic feature learning, and long-term dependencies capture. This is one of the first studies to compare the two widely used deep learning models and demonstrate the superior performance of the RNN model for clinical NER.

Clinical Named Entity Recognition Using Deep Learning Models

PubMed Central

Wu, Yonghui; Jiang, Min; Xu, Jun; Zhi, Degui; Xu, Hua

2017-01-01

Clinical Named Entity Recognition (NER) is a critical natural language processing (NLP) task to extract important concepts (named entities) from clinical narratives. Researchers have extensively investigated machine learning models for clinical NER. Recently, there have been increasing efforts to apply deep learning models to improve the performance of current clinical NER systems. This study examined two popular deep learning architectures, the Convolutional Neural Network (CNN) and the Recurrent Neural Network (RNN), to extract concepts from clinical texts. We compared the two deep neural network architectures with three baseline Conditional Random Fields (CRFs) models and two state-of-the-art clinical NER systems using the i2b2 2010 clinical concept extraction corpus. The evaluation results showed that the RNN model trained with the word embeddings achieved a new state-of-the- art performance (a strict F1 score of 85.94%) for the defined clinical NER task, outperforming the best-reported system that used both manually defined and unsupervised learning features. This study demonstrates the advantage of using deep neural network architectures for clinical concept extraction, including distributed feature representation, automatic feature learning, and long-term dependencies capture. This is one of the first studies to compare the two widely used deep learning models and demonstrate the superior performance of the RNN model for clinical NER. PMID:29854252

Multispectral embedding-based deep neural network for three-dimensional human pose recovery

NASA Astrophysics Data System (ADS)

Yu, Jialin; Sun, Jifeng

2018-01-01

Monocular image-based three-dimensional (3-D) human pose recovery aims to retrieve 3-D poses using the corresponding two-dimensional image features. Therefore, the pose recovery performance highly depends on the image representations. We propose a multispectral embedding-based deep neural network (MSEDNN) to automatically obtain the most discriminative features from multiple deep convolutional neural networks and then embed their penultimate fully connected layers into a low-dimensional manifold. This compact manifold can explore not only the optimum output from multiple deep networks but also the complementary properties of them. Furthermore, the distribution of each hierarchy discriminative manifold is sufficiently smooth so that the training process of our MSEDNN can be effectively implemented only using few labeled data. Our proposed network contains a body joint detector and a human pose regressor that are jointly trained. Extensive experiments conducted on four databases show that our proposed MSEDNN can achieve the best recovery performance compared with the state-of-the-art methods.

Applications of deep convolutional neural networks to digitized natural history collections

PubMed Central

Frandsen, Paul B.; Dikow, Rebecca B.; Brown, Abel; Orli, Sylvia; Peters, Melinda; Metallo, Adam; Funk, Vicki A.; Dorr, Laurence J.

2017-01-01

Abstract Natural history collections contain data that are critical for many scientific endeavors. Recent efforts in mass digitization are generating large datasets from these collections that can provide unprecedented insight. Here, we present examples of how deep convolutional neural networks can be applied in analyses of imaged herbarium specimens. We first demonstrate that a convolutional neural network can detect mercury-stained specimens across a collection with 90% accuracy. We then show that such a network can correctly distinguish two morphologically similar plant families 96% of the time. Discarding the most challenging specimen images increases accuracy to 94% and 99%, respectively. These results highlight the importance of mass digitization and deep learning approaches and reveal how they can together deliver powerful new investigative tools. PMID:29200929

Underwater Inherent Optical Properties Estimation Using a Depth Aided Deep Neural Network.

PubMed

Yu, Zhibin; Wang, Yubo; Zheng, Bing; Zheng, Haiyong; Wang, Nan; Gu, Zhaorui

2017-01-01

Underwater inherent optical properties (IOPs) are the fundamental clues to many research fields such as marine optics, marine biology, and underwater vision. Currently, beam transmissometers and optical sensors are considered as the ideal IOPs measuring methods. But these methods are inflexible and expensive to be deployed. To overcome this problem, we aim to develop a novel measuring method using only a single underwater image with the help of deep artificial neural network. The power of artificial neural network has been proved in image processing and computer vision fields with deep learning technology. However, image-based IOPs estimation is a quite different and challenging task. Unlike the traditional applications such as image classification or localization, IOP estimation looks at the transparency of the water between the camera and the target objects to estimate multiple optical properties simultaneously. In this paper, we propose a novel Depth Aided (DA) deep neural network structure for IOPs estimation based on a single RGB image that is even noisy. The imaging depth information is considered as an aided input to help our model make better decision.

TopologyNet: Topology based deep convolutional and multi-task neural networks for biomolecular property predictions

PubMed Central

2017-01-01

Although deep learning approaches have had tremendous success in image, video and audio processing, computer vision, and speech recognition, their applications to three-dimensional (3D) biomolecular structural data sets have been hindered by the geometric and biological complexity. To address this problem we introduce the element-specific persistent homology (ESPH) method. ESPH represents 3D complex geometry by one-dimensional (1D) topological invariants and retains important biological information via a multichannel image-like representation. This representation reveals hidden structure-function relationships in biomolecules. We further integrate ESPH and deep convolutional neural networks to construct a multichannel topological neural network (TopologyNet) for the predictions of protein-ligand binding affinities and protein stability changes upon mutation. To overcome the deep learning limitations from small and noisy training sets, we propose a multi-task multichannel topological convolutional neural network (MM-TCNN). We demonstrate that TopologyNet outperforms the latest methods in the prediction of protein-ligand binding affinities, mutation induced globular protein folding free energy changes, and mutation induced membrane protein folding free energy changes. Availability: weilab.math.msu.edu/TDL/ PMID:28749969

Prediction of properties of wheat dough using intelligent deep belief networks

NASA Astrophysics Data System (ADS)

Guha, Paramita; Bhatnagar, Taru; Pal, Ishan; Kamboj, Uma; Mishra, Sunita

2017-11-01

In this paper, the rheological and chemical properties of wheat dough are predicted using deep belief networks. Wheat grains are stored at controlled environmental conditions. The internal parameters of grains viz., protein, fat, carbohydrates, moisture, ash are determined using standard chemical analysis and viscosity of the dough is measured using Rheometer. Here, fat, carbohydrates, moisture, ash and temperature are considered as inputs whereas protein and viscosity are chosen as outputs. The prediction algorithm is developed using deep neural network where each layer is trained greedily using restricted Boltzmann machine (RBM) networks. The overall network is finally fine-tuned using standard neural network technique. In most literature, it has been found that fine-tuning is done using back-propagation technique. In this paper, a new algorithm is proposed in which each layer is tuned using RBM and the final network is fine-tuned using deep neural network (DNN). It has been observed that with the proposed algorithm, errors between the actual and predicted outputs are less compared to the conventional algorithm. Hence, the given network can be considered as beneficial as it predicts the outputs more accurately. Numerical results along with discussions are presented.

Student beats the teacher: deep neural networks for lateral ventricles segmentation in brain MR

NASA Astrophysics Data System (ADS)

Ghafoorian, Mohsen; Teuwen, Jonas; Manniesing, Rashindra; Leeuw, Frank-Erik d.; van Ginneken, Bram; Karssemeijer, Nico; Platel, Bram

2018-03-01

Ventricular volume and its progression are known to be linked to several brain diseases such as dementia and schizophrenia. Therefore accurate measurement of ventricle volume is vital for longitudinal studies on these disorders, making automated ventricle segmentation algorithms desirable. In the past few years, deep neural networks have shown to outperform the classical models in many imaging domains. However, the success of deep networks is dependent on manually labeled data sets, which are expensive to acquire especially for higher dimensional data in the medical domain. In this work, we show that deep neural networks can be trained on muchcheaper-to-acquire pseudo-labels (e.g., generated by other automated less accurate methods) and still produce more accurate segmentations compared to the quality of the labels. To show this, we use noisy segmentation labels generated by a conventional region growing algorithm to train a deep network for lateral ventricle segmentation. Then on a large manually annotated test set, we show that the network significantly outperforms the conventional region growing algorithm which was used to produce the training labels for the network. Our experiments report a Dice Similarity Coefficient (DSC) of 0.874 for the trained network compared to 0.754 for the conventional region growing algorithm (p < 0.001).

Processing of chromatic information in a deep convolutional neural network.

PubMed

Flachot, Alban; Gegenfurtner, Karl R

2018-04-01

Deep convolutional neural networks are a class of machine-learning algorithms capable of solving non-trivial tasks, such as object recognition, with human-like performance. Little is known about the exact computations that deep neural networks learn, and to what extent these computations are similar to the ones performed by the primate brain. Here, we investigate how color information is processed in the different layers of the AlexNet deep neural network, originally trained on object classification of over 1.2M images of objects in their natural contexts. We found that the color-responsive units in the first layer of AlexNet learned linear features and were broadly tuned to two directions in color space, analogously to what is known of color responsive cells in the primate thalamus. Moreover, these directions are decorrelated and lead to statistically efficient representations, similar to the cardinal directions of the second-stage color mechanisms in primates. We also found, in analogy to the early stages of the primate visual system, that chromatic and achromatic information were segregated in the early layers of the network. Units in the higher layers of AlexNet exhibit on average a lower responsivity for color than units at earlier stages.

Deep learning in bioinformatics.

PubMed

Min, Seonwoo; Lee, Byunghan; Yoon, Sungroh

2017-09-01

In the era of big data, transformation of biomedical big data into valuable knowledge has been one of the most important challenges in bioinformatics. Deep learning has advanced rapidly since the early 2000s and now demonstrates state-of-the-art performance in various fields. Accordingly, application of deep learning in bioinformatics to gain insight from data has been emphasized in both academia and industry. Here, we review deep learning in bioinformatics, presenting examples of current research. To provide a useful and comprehensive perspective, we categorize research both by the bioinformatics domain (i.e. omics, biomedical imaging, biomedical signal processing) and deep learning architecture (i.e. deep neural networks, convolutional neural networks, recurrent neural networks, emergent architectures) and present brief descriptions of each study. Additionally, we discuss theoretical and practical issues of deep learning in bioinformatics and suggest future research directions. We believe that this review will provide valuable insights and serve as a starting point for researchers to apply deep learning approaches in their bioinformatics studies. © The Author 2016. Published by Oxford University Press. All rights reserved. For Permissions, please email: journals.permissions@oup.com.

A deep convolutional neural network to analyze position averaged convergent beam electron diffraction patterns.

PubMed

Xu, W; LeBeau, J M

2018-05-01

We establish a series of deep convolutional neural networks to automatically analyze position averaged convergent beam electron diffraction patterns. The networks first calibrate the zero-order disk size, center position, and rotation without the need for pretreating the data. With the aligned data, additional networks then measure the sample thickness and tilt. The performance of the network is explored as a function of a variety of variables including thickness, tilt, and dose. A methodology to explore the response of the neural network to various pattern features is also presented. Processing patterns at a rate of  ∼ 0.1 s/pattern, the network is shown to be orders of magnitude faster than a brute force method while maintaining accuracy. The approach is thus suitable for automatically processing big, 4D STEM data. We also discuss the generality of the method to other materials/orientations as well as a hybrid approach that combines the features of the neural network with least squares fitting for even more robust analysis. The source code is available at https://github.com/subangstrom/DeepDiffraction. Copyright © 2018 Elsevier B.V. All rights reserved.

Airplane detection in remote sensing images using convolutional neural networks

NASA Astrophysics Data System (ADS)

Ouyang, Chao; Chen, Zhong; Zhang, Feng; Zhang, Yifei

2018-03-01

Airplane detection in remote sensing images remains a challenging problem and has also been taking a great interest to researchers. In this paper we propose an effective method to detect airplanes in remote sensing images using convolutional neural networks. Deep learning methods show greater advantages than the traditional methods with the rise of deep neural networks in target detection, and we give an explanation why this happens. To improve the performance on detection of airplane, we combine a region proposal algorithm with convolutional neural networks. And in the training phase, we divide the background into multi classes rather than one class, which can reduce false alarms. Our experimental results show that the proposed method is effective and robust in detecting airplane.

Convolutional neural network for road extraction

NASA Astrophysics Data System (ADS)

Li, Junping; Ding, Yazhou; Feng, Fajie; Xiong, Baoyu; Cui, Weihong

2017-11-01

In this paper, the convolution neural network with large block input and small block output was used to extract road. To reflect the complex road characteristics in the study area, a deep convolution neural network VGG19 was conducted for road extraction. Based on the analysis of the characteristics of different sizes of input block, output block and the extraction effect, the votes of deep convolutional neural networks was used as the final road prediction. The study image was from GF-2 panchromatic and multi-spectral fusion in Yinchuan. The precision of road extraction was 91%. The experiments showed that model averaging can improve the accuracy to some extent. At the same time, this paper gave some advice about the choice of input block size and output block size.

Wishart Deep Stacking Network for Fast POLSAR Image Classification.

PubMed

Jiao, Licheng; Liu, Fang

2016-05-11

Inspired by the popular deep learning architecture - Deep Stacking Network (DSN), a specific deep model for polarimetric synthetic aperture radar (POLSAR) image classification is proposed in this paper, which is named as Wishart Deep Stacking Network (W-DSN). First of all, a fast implementation of Wishart distance is achieved by a special linear transformation, which speeds up the classification of POLSAR image and makes it possible to use this polarimetric information in the following Neural Network (NN). Then a single-hidden-layer neural network based on the fast Wishart distance is defined for POLSAR image classification, which is named as Wishart Network (WN) and improves the classification accuracy. Finally, a multi-layer neural network is formed by stacking WNs, which is in fact the proposed deep learning architecture W-DSN for POLSAR image classification and improves the classification accuracy further. In addition, the structure of WN can be expanded in a straightforward way by adding hidden units if necessary, as well as the structure of the W-DSN. As a preliminary exploration on formulating specific deep learning architecture for POLSAR image classification, the proposed methods may establish a simple but clever connection between POLSAR image interpretation and deep learning. The experiment results tested on real POLSAR image show that the fast implementation of Wishart distance is very efficient (a POLSAR image with 768000 pixels can be classified in 0.53s), and both the single-hidden-layer architecture WN and the deep learning architecture W-DSN for POLSAR image classification perform well and work efficiently.

DeepSurv: personalized treatment recommender system using a Cox proportional hazards deep neural network.

PubMed

Katzman, Jared L; Shaham, Uri; Cloninger, Alexander; Bates, Jonathan; Jiang, Tingting; Kluger, Yuval

2018-02-26

Medical practitioners use survival models to explore and understand the relationships between patients' covariates (e.g. clinical and genetic features) and the effectiveness of various treatment options. Standard survival models like the linear Cox proportional hazards model require extensive feature engineering or prior medical knowledge to model treatment interaction at an individual level. While nonlinear survival methods, such as neural networks and survival forests, can inherently model these high-level interaction terms, they have yet to be shown as effective treatment recommender systems. We introduce DeepSurv, a Cox proportional hazards deep neural network and state-of-the-art survival method for modeling interactions between a patient's covariates and treatment effectiveness in order to provide personalized treatment recommendations. We perform a number of experiments training DeepSurv on simulated and real survival data. We demonstrate that DeepSurv performs as well as or better than other state-of-the-art survival models and validate that DeepSurv successfully models increasingly complex relationships between a patient's covariates and their risk of failure. We then show how DeepSurv models the relationship between a patient's features and effectiveness of different treatment options to show how DeepSurv can be used to provide individual treatment recommendations. Finally, we train DeepSurv on real clinical studies to demonstrate how it's personalized treatment recommendations would increase the survival time of a set of patients. The predictive and modeling capabilities of DeepSurv will enable medical researchers to use deep neural networks as a tool in their exploration, understanding, and prediction of the effects of a patient's characteristics on their risk of failure.

Keypoint Density-Based Region Proposal for Fine-Grained Object Detection and Classification Using Regions with Convolutional Neural Network Features

DTIC Science & Technology

2015-12-15

Keypoint Density-based Region Proposal for Fine-Grained Object Detection and Classification using Regions with Convolutional Neural Network ... Convolutional Neural Networks (CNNs) enable them to outperform conventional techniques on standard object detection and classification tasks, their...detection accuracy and speed on the fine-grained Caltech UCSD bird dataset (Wah et al., 2011). Recently, Convolutional Neural Networks (CNNs), a deep

Gas Classification Using Deep Convolutional Neural Networks.

PubMed

Peng, Pai; Zhao, Xiaojin; Pan, Xiaofang; Ye, Wenbin

2018-01-08

In this work, we propose a novel Deep Convolutional Neural Network (DCNN) tailored for gas classification. Inspired by the great success of DCNN in the field of computer vision, we designed a DCNN with up to 38 layers. In general, the proposed gas neural network, named GasNet, consists of: six convolutional blocks, each block consist of six layers; a pooling layer; and a fully-connected layer. Together, these various layers make up a powerful deep model for gas classification. Experimental results show that the proposed DCNN method is an effective technique for classifying electronic nose data. We also demonstrate that the DCNN method can provide higher classification accuracy than comparable Support Vector Machine (SVM) methods and Multiple Layer Perceptron (MLP).

Gas Classification Using Deep Convolutional Neural Networks

PubMed Central

Peng, Pai; Zhao, Xiaojin; Pan, Xiaofang; Ye, Wenbin

2018-01-01

In this work, we propose a novel Deep Convolutional Neural Network (DCNN) tailored for gas classification. Inspired by the great success of DCNN in the field of computer vision, we designed a DCNN with up to 38 layers. In general, the proposed gas neural network, named GasNet, consists of: six convolutional blocks, each block consist of six layers; a pooling layer; and a fully-connected layer. Together, these various layers make up a powerful deep model for gas classification. Experimental results show that the proposed DCNN method is an effective technique for classifying electronic nose data. We also demonstrate that the DCNN method can provide higher classification accuracy than comparable Support Vector Machine (SVM) methods and Multiple Layer Perceptron (MLP). PMID:29316723

The Role of Architectural and Learning Constraints in Neural Network Models: A Case Study on Visual Space Coding.

PubMed

Testolin, Alberto; De Filippo De Grazia, Michele; Zorzi, Marco

2017-01-01

The recent "deep learning revolution" in artificial neural networks had strong impact and widespread deployment for engineering applications, but the use of deep learning for neurocomputational modeling has been so far limited. In this article we argue that unsupervised deep learning represents an important step forward for improving neurocomputational models of perception and cognition, because it emphasizes the role of generative learning as opposed to discriminative (supervised) learning. As a case study, we present a series of simulations investigating the emergence of neural coding of visual space for sensorimotor transformations. We compare different network architectures commonly used as building blocks for unsupervised deep learning by systematically testing the type of receptive fields and gain modulation developed by the hidden neurons. In particular, we compare Restricted Boltzmann Machines (RBMs), which are stochastic, generative networks with bidirectional connections trained using contrastive divergence, with autoencoders, which are deterministic networks trained using error backpropagation. For both learning architectures we also explore the role of sparse coding, which has been identified as a fundamental principle of neural computation. The unsupervised models are then compared with supervised, feed-forward networks that learn an explicit mapping between different spatial reference frames. Our simulations show that both architectural and learning constraints strongly influenced the emergent coding of visual space in terms of distribution of tuning functions at the level of single neurons. Unsupervised models, and particularly RBMs, were found to more closely adhere to neurophysiological data from single-cell recordings in the primate parietal cortex. These results provide new insights into how basic properties of artificial neural networks might be relevant for modeling neural information processing in biological systems.

The Role of Architectural and Learning Constraints in Neural Network Models: A Case Study on Visual Space Coding

PubMed Central

Testolin, Alberto; De Filippo De Grazia, Michele; Zorzi, Marco

2017-01-01

The recent “deep learning revolution” in artificial neural networks had strong impact and widespread deployment for engineering applications, but the use of deep learning for neurocomputational modeling has been so far limited. In this article we argue that unsupervised deep learning represents an important step forward for improving neurocomputational models of perception and cognition, because it emphasizes the role of generative learning as opposed to discriminative (supervised) learning. As a case study, we present a series of simulations investigating the emergence of neural coding of visual space for sensorimotor transformations. We compare different network architectures commonly used as building blocks for unsupervised deep learning by systematically testing the type of receptive fields and gain modulation developed by the hidden neurons. In particular, we compare Restricted Boltzmann Machines (RBMs), which are stochastic, generative networks with bidirectional connections trained using contrastive divergence, with autoencoders, which are deterministic networks trained using error backpropagation. For both learning architectures we also explore the role of sparse coding, which has been identified as a fundamental principle of neural computation. The unsupervised models are then compared with supervised, feed-forward networks that learn an explicit mapping between different spatial reference frames. Our simulations show that both architectural and learning constraints strongly influenced the emergent coding of visual space in terms of distribution of tuning functions at the level of single neurons. Unsupervised models, and particularly RBMs, were found to more closely adhere to neurophysiological data from single-cell recordings in the primate parietal cortex. These results provide new insights into how basic properties of artificial neural networks might be relevant for modeling neural information processing in biological systems. PMID:28377709

Learning representations for the early detection of sepsis with deep neural networks.

PubMed

Kam, Hye Jin; Kim, Ha Young

2017-10-01

Sepsis is one of the leading causes of death in intensive care unit patients. Early detection of sepsis is vital because mortality increases as the sepsis stage worsens. This study aimed to develop detection models for the early stage of sepsis using deep learning methodologies, and to compare the feasibility and performance of the new deep learning methodology with those of the regression method with conventional temporal feature extraction. Study group selection adhered to the InSight model. The results of the deep learning-based models and the InSight model were compared. With deep feedforward networks, the area under the ROC curve (AUC) of the models were 0.887 and 0.915 for the InSight and the new feature sets, respectively. For the model with the combined feature set, the AUC was the same as that of the basic feature set (0.915). For the long short-term memory model, only the basic feature set was applied and the AUC improved to 0.929 compared with the existing 0.887 of the InSight model. The contributions of this paper can be summarized in three ways: (i) improved performance without feature extraction using domain knowledge, (ii) verification of feature extraction capability of deep neural networks through comparison with reference features, and (iii) improved performance with feedforward neural networks using long short-term memory, a neural network architecture that can learn sequential patterns. Copyright © 2017 Elsevier Ltd. All rights reserved.

Magnetoencephalographic imaging of deep corticostriatal network activity during a rewards paradigm.

PubMed

Kanal, Eliezer Y; Sun, Mingui; Ozkurt, Tolga E; Jia, Wenyan; Sclabassi, Robert

2009-01-01

The human rewards network is a complex system spanning both cortical and subcortical regions. While much is known about the functions of the various components of the network, research on the behavior of the network as a whole has been stymied due to an inability to detect signals at a high enough temporal resolution from both superficial and deep network components simultaneously. In this paper, we describe the application of magnetoencephalographic imaging (MEG) combined with advanced signal processing techniques to this problem. Using data collected while subjects performed a rewards-related gambling paradigm demonstrated to activate the rewards network, we were able to identify neural signals which correspond to deep network activity. We also show that this signal was not observable prior to filtration. These results suggest that MEG imaging may be a viable tool for the detection of deep neural activity.

ShapeShop: Towards Understanding Deep Learning Representations via Interactive Experimentation.

PubMed

Hohman, Fred; Hodas, Nathan; Chau, Duen Horng

2017-05-01

Deep learning is the driving force behind many recent technologies; however, deep neural networks are often viewed as "black-boxes" due to their internal complexity that is hard to understand. Little research focuses on helping people explore and understand the relationship between a user's data and the learned representations in deep learning models. We present our ongoing work, ShapeShop, an interactive system for visualizing and understanding what semantics a neural network model has learned. Built using standard web technologies, ShapeShop allows users to experiment with and compare deep learning models to help explore the robustness of image classifiers.

Pulmonary Nodule Classification with Deep Convolutional Neural Networks on Computed Tomography Images.

PubMed

Li, Wei; Cao, Peng; Zhao, Dazhe; Wang, Junbo

2016-01-01

Computer aided detection (CAD) systems can assist radiologists by offering a second opinion on early diagnosis of lung cancer. Classification and feature representation play critical roles in false-positive reduction (FPR) in lung nodule CAD. We design a deep convolutional neural networks method for nodule classification, which has an advantage of autolearning representation and strong generalization ability. A specified network structure for nodule images is proposed to solve the recognition of three types of nodules, that is, solid, semisolid, and ground glass opacity (GGO). Deep convolutional neural networks are trained by 62,492 regions-of-interest (ROIs) samples including 40,772 nodules and 21,720 nonnodules from the Lung Image Database Consortium (LIDC) database. Experimental results demonstrate the effectiveness of the proposed method in terms of sensitivity and overall accuracy and that it consistently outperforms the competing methods.

Human fatigue expression recognition through image-based dynamic multi-information and bimodal deep learning

NASA Astrophysics Data System (ADS)

Zhao, Lei; Wang, Zengcai; Wang, Xiaojin; Qi, Yazhou; Liu, Qing; Zhang, Guoxin

2016-09-01

Human fatigue is an important cause of traffic accidents. To improve the safety of transportation, we propose, in this paper, a framework for fatigue expression recognition using image-based facial dynamic multi-information and a bimodal deep neural network. First, the landmark of face region and the texture of eye region, which complement each other in fatigue expression recognition, are extracted from facial image sequences captured by a single camera. Then, two stacked autoencoder neural networks are trained for landmark and texture, respectively. Finally, the two trained neural networks are combined by learning a joint layer on top of them to construct a bimodal deep neural network. The model can be used to extract a unified representation that fuses landmark and texture modalities together and classify fatigue expressions accurately. The proposed system is tested on a human fatigue dataset obtained from an actual driving environment. The experimental results demonstrate that the proposed method performs stably and robustly, and that the average accuracy achieves 96.2%.

Evaluation of Deep Learning Models for Predicting CO2 Flux

NASA Astrophysics Data System (ADS)

Halem, M.; Nguyen, P.; Frankel, D.

2017-12-01

Artificial neural networks have been employed to calculate surface flux measurements from station data because they are able to fit highly nonlinear relations between input and output variables without knowing the detail relationships between the variables. However, the accuracy in performing neural net estimates of CO2 flux from observations of CO2 and other atmospheric variables is influenced by the architecture of the neural model, the availability, and complexity of interactions between physical variables such as wind, temperature, and indirect variables like latent heat, and sensible heat, etc. We evaluate two deep learning models, feed forward and recurrent neural network models to learn how they each respond to the physical measurements, time dependency of the measurements of CO2 concentration, humidity, pressure, temperature, wind speed etc. for predicting the CO2 flux. In this paper, we focus on a) building neural network models for estimating CO2 flux based on DOE data from tower Atmospheric Radiation Measurement data; b) evaluating the impact of choosing the surface variables and model hyper-parameters on the accuracy and predictions of surface flux; c) assessing the applicability of the neural network models on estimate CO2 flux by using OCO-2 satellite data; d) studying the efficiency of using GPU-acceleration for neural network performance using IBM Power AI deep learning software and packages on IBM Minsky system.

Dynamics of scene representations in the human brain revealed by magnetoencephalography and deep neural networks

PubMed Central

Cichy, Radoslaw Martin; Khosla, Aditya; Pantazis, Dimitrios; Oliva, Aude

2017-01-01

Human scene recognition is a rapid multistep process evolving over time from single scene image to spatial layout processing. We used multivariate pattern analyses on magnetoencephalography (MEG) data to unravel the time course of this cortical process. Following an early signal for lower-level visual analysis of single scenes at ~100 ms, we found a marker of real-world scene size, i.e. spatial layout processing, at ~250 ms indexing neural representations robust to changes in unrelated scene properties and viewing conditions. For a quantitative model of how scene size representations may arise in the brain, we compared MEG data to a deep neural network model trained on scene classification. Representations of scene size emerged intrinsically in the model, and resolved emerging neural scene size representation. Together our data provide a first description of an electrophysiological signal for layout processing in humans, and suggest that deep neural networks are a promising framework to investigate how spatial layout representations emerge in the human brain. PMID:27039703

Toolkits and Libraries for Deep Learning.

PubMed

Erickson, Bradley J; Korfiatis, Panagiotis; Akkus, Zeynettin; Kline, Timothy; Philbrick, Kenneth

2017-08-01

Deep learning is an important new area of machine learning which encompasses a wide range of neural network architectures designed to complete various tasks. In the medical imaging domain, example tasks include organ segmentation, lesion detection, and tumor classification. The most popular network architecture for deep learning for images is the convolutional neural network (CNN). Whereas traditional machine learning requires determination and calculation of features from which the algorithm learns, deep learning approaches learn the important features as well as the proper weighting of those features to make predictions for new data. In this paper, we will describe some of the libraries and tools that are available to aid in the construction and efficient execution of deep learning as applied to medical images.

Cough event classification by pretrained deep neural network.

PubMed

Liu, Jia-Ming; You, Mingyu; Wang, Zheng; Li, Guo-Zheng; Xu, Xianghuai; Qiu, Zhongmin

2015-01-01

Cough is an essential symptom in respiratory diseases. In the measurement of cough severity, an accurate and objective cough monitor is expected by respiratory disease society. This paper aims to introduce a better performed algorithm, pretrained deep neural network (DNN), to the cough classification problem, which is a key step in the cough monitor. The deep neural network models are built from two steps, pretrain and fine-tuning, followed by a Hidden Markov Model (HMM) decoder to capture tamporal information of the audio signals. By unsupervised pretraining a deep belief network, a good initialization for a deep neural network is learned. Then the fine-tuning step is a back propogation tuning the neural network so that it can predict the observation probability associated with each HMM states, where the HMM states are originally achieved by force-alignment with a Gaussian Mixture Model Hidden Markov Model (GMM-HMM) on the training samples. Three cough HMMs and one noncough HMM are employed to model coughs and noncoughs respectively. The final decision is made based on viterbi decoding algorihtm that generates the most likely HMM sequence for each sample. A sample is labeled as cough if a cough HMM is found in the sequence. The experiments were conducted on a dataset that was collected from 22 patients with respiratory diseases. Patient dependent (PD) and patient independent (PI) experimental settings were used to evaluate the models. Five criteria, sensitivity, specificity, F1, macro average and micro average are shown to depict different aspects of the models. From overall evaluation criteria, the DNN based methods are superior to traditional GMM-HMM based method on F1 and micro average with maximal 14% and 11% error reduction in PD and 7% and 10% in PI, meanwhile keep similar performances on macro average. They also surpass GMM-HMM model on specificity with maximal 14% error reduction on both PD and PI. In this paper, we tried pretrained deep neural network in cough classification problem. Our results showed that comparing with the conventional GMM-HMM framework, the HMM-DNN could get better overall performance on cough classification task.

Deep Learning: A Primer for Radiologists.

PubMed

Chartrand, Gabriel; Cheng, Phillip M; Vorontsov, Eugene; Drozdzal, Michal; Turcotte, Simon; Pal, Christopher J; Kadoury, Samuel; Tang, An

2017-01-01

Deep learning is a class of machine learning methods that are gaining success and attracting interest in many domains, including computer vision, speech recognition, natural language processing, and playing games. Deep learning methods produce a mapping from raw inputs to desired outputs (eg, image classes). Unlike traditional machine learning methods, which require hand-engineered feature extraction from inputs, deep learning methods learn these features directly from data. With the advent of large datasets and increased computing power, these methods can produce models with exceptional performance. These models are multilayer artificial neural networks, loosely inspired by biologic neural systems. Weighted connections between nodes (neurons) in the network are iteratively adjusted based on example pairs of inputs and target outputs by back-propagating a corrective error signal through the network. For computer vision tasks, convolutional neural networks (CNNs) have proven to be effective. Recently, several clinical applications of CNNs have been proposed and studied in radiology for classification, detection, and segmentation tasks. This article reviews the key concepts of deep learning for clinical radiologists, discusses technical requirements, describes emerging applications in clinical radiology, and outlines limitations and future directions in this field. Radiologists should become familiar with the principles and potential applications of deep learning in medical imaging. © RSNA, 2017.

Deep convolutional neural network based antenna selection in multiple-input multiple-output system

NASA Astrophysics Data System (ADS)

Cai, Jiaxin; Li, Yan; Hu, Ying

2018-03-01

Antenna selection of wireless communication system has attracted increasing attention due to the challenge of keeping a balance between communication performance and computational complexity in large-scale Multiple-Input MultipleOutput antenna systems. Recently, deep learning based methods have achieved promising performance for large-scale data processing and analysis in many application fields. This paper is the first attempt to introduce the deep learning technique into the field of Multiple-Input Multiple-Output antenna selection in wireless communications. First, the label of attenuation coefficients channel matrix is generated by minimizing the key performance indicator of training antenna systems. Then, a deep convolutional neural network that explicitly exploits the massive latent cues of attenuation coefficients is learned on the training antenna systems. Finally, we use the adopted deep convolutional neural network to classify the channel matrix labels of test antennas and select the optimal antenna subset. Simulation experimental results demonstrate that our method can achieve better performance than the state-of-the-art baselines for data-driven based wireless antenna selection.

Gait Recognition Based on Convolutional Neural Networks

NASA Astrophysics Data System (ADS)

Sokolova, A.; Konushin, A.

2017-05-01

In this work we investigate the problem of people recognition by their gait. For this task, we implement deep learning approach using the optical flow as the main source of motion information and combine neural feature extraction with the additional embedding of descriptors for representation improvement. In order to find the best heuristics, we compare several deep neural network architectures, learning and classification strategies. The experiments were made on two popular datasets for gait recognition, so we investigate their advantages and disadvantages and the transferability of considered methods.

Hourly air pollution concentrations and their important predictors over Houston, Texas using deep neural networks: case study of DISCOVER-AQ time period

NASA Astrophysics Data System (ADS)

Eslami, E.; Choi, Y.; Roy, A.

2017-12-01

Air quality forecasting carried out by chemical transport models often show significant error. This study uses a deep-learning approach over the Houston-Galveston-Brazoria (HGB) area to overcome this forecasting challenge, for the DISCOVER-AQ period (September 2013). Two approaches, deep neural network (DNN) using a Multi-Layer Perceptron (MLP) and Restricted Boltzmann Machine (RBM) were utilized. The proposed approaches analyzed input data by identifying features abstracted from its previous layer using a stepwise method. The approaches predicted hourly ozone and PM in September 2013 using several predictors of prior three days, including wind fields, temperature, relative humidity, cloud fraction, precipitation along with PM, ozone, and NOx concentrations. Model-measurement comparisons for available monitoring sites reported Indexes of Agreement (IOA) of around 0.95 for both DNN and RBM. A standard artificial neural network (ANN) (IOA=0.90) with similar architecture showed poorer performance than the deep networks, clearly demonstrating the superiority of the deep approaches. Additionally, each network (both deep and standard) performed significantly better than a previous CMAQ study, which showed an IOA of less than 0.80. The most influential input variables were identified using their associated weights, which represented the sensitivity of ozone to input parameters. The results indicate deep learning approaches can achieve more accurate ozone forecasting and identify the important input variables for ozone predictions in metropolitan areas.

De novo peptide sequencing by deep learning

PubMed Central

Tran, Ngoc Hieu; Zhang, Xianglilan; Xin, Lei; Shan, Baozhen; Li, Ming

2017-01-01

De novo peptide sequencing from tandem MS data is the key technology in proteomics for the characterization of proteins, especially for new sequences, such as mAbs. In this study, we propose a deep neural network model, DeepNovo, for de novo peptide sequencing. DeepNovo architecture combines recent advances in convolutional neural networks and recurrent neural networks to learn features of tandem mass spectra, fragment ions, and sequence patterns of peptides. The networks are further integrated with local dynamic programming to solve the complex optimization task of de novo sequencing. We evaluated the method on a wide variety of species and found that DeepNovo considerably outperformed state of the art methods, achieving 7.7–22.9% higher accuracy at the amino acid level and 38.1–64.0% higher accuracy at the peptide level. We further used DeepNovo to automatically reconstruct the complete sequences of antibody light and heavy chains of mouse, achieving 97.5–100% coverage and 97.2–99.5% accuracy, without assisting databases. Moreover, DeepNovo is retrainable to adapt to any sources of data and provides a complete end-to-end training and prediction solution to the de novo sequencing problem. Not only does our study extend the deep learning revolution to a new field, but it also shows an innovative approach in solving optimization problems by using deep learning and dynamic programming. PMID:28720701

A Deep Learning Network Approach to ab initio Protein Secondary Structure Prediction

PubMed Central

Spencer, Matt; Eickholt, Jesse; Cheng, Jianlin

2014-01-01

Ab initio protein secondary structure (SS) predictions are utilized to generate tertiary structure predictions, which are increasingly demanded due to the rapid discovery of proteins. Although recent developments have slightly exceeded previous methods of SS prediction, accuracy has stagnated around 80% and many wonder if prediction cannot be advanced beyond this ceiling. Disciplines that have traditionally employed neural networks are experimenting with novel deep learning techniques in attempts to stimulate progress. Since neural networks have historically played an important role in SS prediction, we wanted to determine whether deep learning could contribute to the advancement of this field as well. We developed an SS predictor that makes use of the position-specific scoring matrix generated by PSI-BLAST and deep learning network architectures, which we call DNSS. Graphical processing units and CUDA software optimize the deep network architecture and efficiently train the deep networks. Optimal parameters for the training process were determined, and a workflow comprising three separately trained deep networks was constructed in order to make refined predictions. This deep learning network approach was used to predict SS for a fully independent test data set of 198 proteins, achieving a Q3 accuracy of 80.7% and a Sov accuracy of 74.2%. PMID:25750595

A Deep Learning Network Approach to ab initio Protein Secondary Structure Prediction.

PubMed

Spencer, Matt; Eickholt, Jesse; Jianlin Cheng

2015-01-01

Ab initio protein secondary structure (SS) predictions are utilized to generate tertiary structure predictions, which are increasingly demanded due to the rapid discovery of proteins. Although recent developments have slightly exceeded previous methods of SS prediction, accuracy has stagnated around 80 percent and many wonder if prediction cannot be advanced beyond this ceiling. Disciplines that have traditionally employed neural networks are experimenting with novel deep learning techniques in attempts to stimulate progress. Since neural networks have historically played an important role in SS prediction, we wanted to determine whether deep learning could contribute to the advancement of this field as well. We developed an SS predictor that makes use of the position-specific scoring matrix generated by PSI-BLAST and deep learning network architectures, which we call DNSS. Graphical processing units and CUDA software optimize the deep network architecture and efficiently train the deep networks. Optimal parameters for the training process were determined, and a workflow comprising three separately trained deep networks was constructed in order to make refined predictions. This deep learning network approach was used to predict SS for a fully independent test dataset of 198 proteins, achieving a Q3 accuracy of 80.7 percent and a Sov accuracy of 74.2 percent.

Spatiotemporal Recurrent Convolutional Networks for Traffic Prediction in Transportation Networks

PubMed Central

Yu, Haiyang; Wu, Zhihai; Wang, Shuqin; Wang, Yunpeng; Ma, Xiaolei

2017-01-01

Predicting large-scale transportation network traffic has become an important and challenging topic in recent decades. Inspired by the domain knowledge of motion prediction, in which the future motion of an object can be predicted based on previous scenes, we propose a network grid representation method that can retain the fine-scale structure of a transportation network. Network-wide traffic speeds are converted into a series of static images and input into a novel deep architecture, namely, spatiotemporal recurrent convolutional networks (SRCNs), for traffic forecasting. The proposed SRCNs inherit the advantages of deep convolutional neural networks (DCNNs) and long short-term memory (LSTM) neural networks. The spatial dependencies of network-wide traffic can be captured by DCNNs, and the temporal dynamics can be learned by LSTMs. An experiment on a Beijing transportation network with 278 links demonstrates that SRCNs outperform other deep learning-based algorithms in both short-term and long-term traffic prediction. PMID:28672867

Spatiotemporal Recurrent Convolutional Networks for Traffic Prediction in Transportation Networks.

PubMed

Yu, Haiyang; Wu, Zhihai; Wang, Shuqin; Wang, Yunpeng; Ma, Xiaolei

2017-06-26

Predicting large-scale transportation network traffic has become an important and challenging topic in recent decades. Inspired by the domain knowledge of motion prediction, in which the future motion of an object can be predicted based on previous scenes, we propose a network grid representation method that can retain the fine-scale structure of a transportation network. Network-wide traffic speeds are converted into a series of static images and input into a novel deep architecture, namely, spatiotemporal recurrent convolutional networks (SRCNs), for traffic forecasting. The proposed SRCNs inherit the advantages of deep convolutional neural networks (DCNNs) and long short-term memory (LSTM) neural networks. The spatial dependencies of network-wide traffic can be captured by DCNNs, and the temporal dynamics can be learned by LSTMs. An experiment on a Beijing transportation network with 278 links demonstrates that SRCNs outperform other deep learning-based algorithms in both short-term and long-term traffic prediction.

The applications of deep neural networks to sdBV classification

NASA Astrophysics Data System (ADS)

Boudreaux, Thomas M.

2017-12-01

With several new large-scale surveys on the horizon, including LSST, TESS, ZTF, and Evryscope, faster and more accurate analysis methods will be required to adequately process the enormous amount of data produced. Deep learning, used in industry for years now, allows for advanced feature detection in minimally prepared datasets at very high speeds; however, despite the advantages of this method, its application to astrophysics has not yet been extensively explored. This dearth may be due to a lack of training data available to researchers. Here we generate synthetic data loosely mimicking the properties of acoustic mode pulsating stars and we show that two separate paradigms of deep learning - the Artificial Neural Network And the Convolutional Neural Network - can both be used to classify this synthetic data effectively. And that additionally this classification can be performed at relatively high levels of accuracy with minimal time spent adjusting network hyperparameters.

Shakeout: A New Approach to Regularized Deep Neural Network Training.

PubMed

Kang, Guoliang; Li, Jun; Tao, Dacheng

2018-05-01

Recent years have witnessed the success of deep neural networks in dealing with a plenty of practical problems. Dropout has played an essential role in many successful deep neural networks, by inducing regularization in the model training. In this paper, we present a new regularized training approach: Shakeout. Instead of randomly discarding units as Dropout does at the training stage, Shakeout randomly chooses to enhance or reverse each unit's contribution to the next layer. This minor modification of Dropout has the statistical trait: the regularizer induced by Shakeout adaptively combines , and regularization terms. Our classification experiments with representative deep architectures on image datasets MNIST, CIFAR-10 and ImageNet show that Shakeout deals with over-fitting effectively and outperforms Dropout. We empirically demonstrate that Shakeout leads to sparser weights under both unsupervised and supervised settings. Shakeout also leads to the grouping effect of the input units in a layer. Considering the weights in reflecting the importance of connections, Shakeout is superior to Dropout, which is valuable for the deep model compression. Moreover, we demonstrate that Shakeout can effectively reduce the instability of the training process of the deep architecture.

Learning Traffic as Images: A Deep Convolutional Neural Network for Large-Scale Transportation Network Speed Prediction.

PubMed

Ma, Xiaolei; Dai, Zhuang; He, Zhengbing; Ma, Jihui; Wang, Yong; Wang, Yunpeng

2017-04-10

This paper proposes a convolutional neural network (CNN)-based method that learns traffic as images and predicts large-scale, network-wide traffic speed with a high accuracy. Spatiotemporal traffic dynamics are converted to images describing the time and space relations of traffic flow via a two-dimensional time-space matrix. A CNN is applied to the image following two consecutive steps: abstract traffic feature extraction and network-wide traffic speed prediction. The effectiveness of the proposed method is evaluated by taking two real-world transportation networks, the second ring road and north-east transportation network in Beijing, as examples, and comparing the method with four prevailing algorithms, namely, ordinary least squares, k-nearest neighbors, artificial neural network, and random forest, and three deep learning architectures, namely, stacked autoencoder, recurrent neural network, and long-short-term memory network. The results show that the proposed method outperforms other algorithms by an average accuracy improvement of 42.91% within an acceptable execution time. The CNN can train the model in a reasonable time and, thus, is suitable for large-scale transportation networks.

Learning Traffic as Images: A Deep Convolutional Neural Network for Large-Scale Transportation Network Speed Prediction

PubMed Central

Ma, Xiaolei; Dai, Zhuang; He, Zhengbing; Ma, Jihui; Wang, Yong; Wang, Yunpeng

2017-01-01

This paper proposes a convolutional neural network (CNN)-based method that learns traffic as images and predicts large-scale, network-wide traffic speed with a high accuracy. Spatiotemporal traffic dynamics are converted to images describing the time and space relations of traffic flow via a two-dimensional time-space matrix. A CNN is applied to the image following two consecutive steps: abstract traffic feature extraction and network-wide traffic speed prediction. The effectiveness of the proposed method is evaluated by taking two real-world transportation networks, the second ring road and north-east transportation network in Beijing, as examples, and comparing the method with four prevailing algorithms, namely, ordinary least squares, k-nearest neighbors, artificial neural network, and random forest, and three deep learning architectures, namely, stacked autoencoder, recurrent neural network, and long-short-term memory network. The results show that the proposed method outperforms other algorithms by an average accuracy improvement of 42.91% within an acceptable execution time. The CNN can train the model in a reasonable time and, thus, is suitable for large-scale transportation networks. PMID:28394270

ShapeShop: Towards Understanding Deep Learning Representations via Interactive Experimentation

DOE Office of Scientific and Technical Information (OSTI.GOV)

Hohman, Frederick M.; Hodas, Nathan O.; Chau, Duen Horng

Deep learning is the driving force behind many recent technologies; however, deep neural networks are often viewed as “black-boxes” due to their internal complexity that is hard to understand. Little research focuses on helping people explore and understand the relationship between a user’s data and the learned representations in deep learning models. We present our ongoing work, ShapeShop, an interactive system for visualizing and understanding what semantics a neural network model has learned. Built using standard web technologies, ShapeShop allows users to experiment with and compare deep learning models to help explore the robustness of image classifiers.

A novel end-to-end classifier using domain transferred deep convolutional neural networks for biomedical images.

PubMed

Pang, Shuchao; Yu, Zhezhou; Orgun, Mehmet A

2017-03-01

Highly accurate classification of biomedical images is an essential task in the clinical diagnosis of numerous medical diseases identified from those images. Traditional image classification methods combined with hand-crafted image feature descriptors and various classifiers are not able to effectively improve the accuracy rate and meet the high requirements of classification of biomedical images. The same also holds true for artificial neural network models directly trained with limited biomedical images used as training data or directly used as a black box to extract the deep features based on another distant dataset. In this study, we propose a highly reliable and accurate end-to-end classifier for all kinds of biomedical images via deep learning and transfer learning. We first apply domain transferred deep convolutional neural network for building a deep model; and then develop an overall deep learning architecture based on the raw pixels of original biomedical images using supervised training. In our model, we do not need the manual design of the feature space, seek an effective feature vector classifier or segment specific detection object and image patches, which are the main technological difficulties in the adoption of traditional image classification methods. Moreover, we do not need to be concerned with whether there are large training sets of annotated biomedical images, affordable parallel computing resources featuring GPUs or long times to wait for training a perfect deep model, which are the main problems to train deep neural networks for biomedical image classification as observed in recent works. With the utilization of a simple data augmentation method and fast convergence speed, our algorithm can achieve the best accuracy rate and outstanding classification ability for biomedical images. We have evaluated our classifier on several well-known public biomedical datasets and compared it with several state-of-the-art approaches. We propose a robust automated end-to-end classifier for biomedical images based on a domain transferred deep convolutional neural network model that shows a highly reliable and accurate performance which has been confirmed on several public biomedical image datasets. Copyright © 2017 Elsevier Ireland Ltd. All rights reserved.

Corpus callosum segmentation using deep neural networks with prior information from multi-atlas images

NASA Astrophysics Data System (ADS)

Park, Gilsoon; Hong, Jinwoo; Lee, Jong-Min

2018-03-01

In human brain, Corpus Callosum (CC) is the largest white matter structure, connecting between right and left hemispheres. Structural features such as shape and size of CC in midsagittal plane are of great significance for analyzing various neurological diseases, for example Alzheimer's disease, autism and epilepsy. For quantitative and qualitative studies of CC in brain MR images, robust segmentation of CC is important. In this paper, we present a novel method for CC segmentation. Our approach is based on deep neural networks and the prior information generated from multi-atlas images. Deep neural networks have recently shown good performance in various image processing field. Convolutional neural networks (CNN) have shown outstanding performance for classification and segmentation in medical image fields. We used convolutional neural networks for CC segmentation. Multi-atlas based segmentation model have been widely used in medical image segmentation because atlas has powerful information about the target structure we want to segment, consisting of MR images and corresponding manual segmentation of the target structure. We combined the prior information, such as location and intensity distribution of target structure (i.e. CC), made from multi-atlas images in CNN training process for more improving training. The CNN with prior information showed better segmentation performance than without.

Decoding of Human Movements Based on Deep Brain Local Field Potentials Using Ensemble Neural Networks

PubMed Central

2017-01-01

Decoding neural activities related to voluntary and involuntary movements is fundamental to understanding human brain motor circuits and neuromotor disorders and can lead to the development of neuromotor prosthetic devices for neurorehabilitation. This study explores using recorded deep brain local field potentials (LFPs) for robust movement decoding of Parkinson's disease (PD) and Dystonia patients. The LFP data from voluntary movement activities such as left and right hand index finger clicking were recorded from patients who underwent surgeries for implantation of deep brain stimulation electrodes. Movement-related LFP signal features were extracted by computing instantaneous power related to motor response in different neural frequency bands. An innovative neural network ensemble classifier has been proposed and developed for accurate prediction of finger movement and its forthcoming laterality. The ensemble classifier contains three base neural network classifiers, namely, feedforward, radial basis, and probabilistic neural networks. The majority voting rule is used to fuse the decisions of the three base classifiers to generate the final decision of the ensemble classifier. The overall decoding performance reaches a level of agreement (kappa value) at about 0.729 ± 0.16 for decoding movement from the resting state and about 0.671 ± 0.14 for decoding left and right visually cued movements. PMID:29201041

Nonparametric Representations for Integrated Inference, Control, and Sensing

DTIC Science & Technology

2015-10-01

Learning (ICML), 2013. [20] Jeff Donahue, Yangqing Jia, Oriol Vinyals, Judy Hoffman, Ning Zhang, Eric Tzeng, and Trevor Darrell. DeCAF: A deep ...unlimited. Multi-layer feature learning “SuperVision” Convolutional Neural Network (CNN) ImageNet Classification with Deep Convolutional Neural Networks...to develop a new framework for autonomous operations that will extend the state of the art in distributed learning and modeling from data, and

Artificial intelligence: Deep neural reasoning

NASA Astrophysics Data System (ADS)

Jaeger, Herbert

2016-10-01

The human brain can solve highly abstract reasoning problems using a neural network that is entirely physical. The underlying mechanisms are only partially understood, but an artificial network provides valuable insight. See Article p.471

Generating Seismograms with Deep Neural Networks

NASA Astrophysics Data System (ADS)

Krischer, L.; Fichtner, A.

2017-12-01

The recent surge of successful uses of deep neural networks in computer vision, speech recognition, and natural language processing, mainly enabled by the availability of fast GPUs and extremely large data sets, is starting to see many applications across all natural sciences. In seismology these are largely confined to classification and discrimination tasks. In this contribution we explore the use of deep neural networks for another class of problems: so called generative models.Generative modelling is a branch of statistics concerned with generating new observed data samples, usually by drawing from some underlying probability distribution. Samples with specific attributes can be generated by conditioning on input variables. In this work we condition on seismic source (mechanism and location) and receiver (location) parameters to generate multi-component seismograms.The deep neural networks are trained on synthetic data calculated with Instaseis (http://instaseis.net, van Driel et al. (2015)) and waveforms from the global ShakeMovie project (http://global.shakemovie.princeton.edu, Tromp et al. (2010)). The underlying radially symmetric or smoothly three dimensional Earth structures result in comparatively small waveform differences from similar events or at close receivers and the networks learn to interpolate between training data samples.Of particular importance is the chosen misfit functional. Generative adversarial networks (Goodfellow et al. (2014)) implement a system in which two networks compete: the generator network creates samples and the discriminator network distinguishes these from the true training examples. Both are trained in an adversarial fashion until the discriminator can no longer distinguish between generated and real samples. We show how this can be applied to seismograms and in particular how it compares to networks trained with more conventional misfit metrics. Last but not least we attempt to shed some light on the black-box nature of neural networks by estimating the quality and uncertainties of the generated seismograms.

Adaptive template generation for amyloid PET using a deep learning approach.

PubMed

Kang, Seung Kwan; Seo, Seongho; Shin, Seong A; Byun, Min Soo; Lee, Dong Young; Kim, Yu Kyeong; Lee, Dong Soo; Lee, Jae Sung

2018-05-11

Accurate spatial normalization (SN) of amyloid positron emission tomography (PET) images for Alzheimer's disease assessment without coregistered anatomical magnetic resonance imaging (MRI) of the same individual is technically challenging. In this study, we applied deep neural networks to generate individually adaptive PET templates for robust and accurate SN of amyloid PET without using matched 3D MR images. Using 681 pairs of simultaneously acquired 11 C-PIB PET and T1-weighted 3D MRI scans of AD, MCI, and cognitively normal subjects, we trained and tested two deep neural networks [convolutional auto-encoder (CAE) and generative adversarial network (GAN)] that produce adaptive best PET templates. More specifically, the networks were trained using 685,100 pieces of augmented data generated by rotating 527 randomly selected datasets and validated using 154 datasets. The input to the supervised neural networks was the 3D PET volume in native space and the label was the spatially normalized 3D PET image using the transformation parameters obtained from MRI-based SN. The proposed deep learning approach significantly enhanced the quantitative accuracy of MRI-less amyloid PET assessment by reducing the SN error observed when an average amyloid PET template is used. Given an input image, the trained deep neural networks rapidly provide individually adaptive 3D PET templates without any discontinuity between the slices (in 0.02 s). As the proposed method does not require 3D MRI for the SN of PET images, it has great potential for use in routine analysis of amyloid PET images in clinical practice and research. © 2018 Wiley Periodicals, Inc.

Beyond the hype: deep neural networks outperform established methods using a ChEMBL bioactivity benchmark set.

PubMed

Lenselink, Eelke B; Ten Dijke, Niels; Bongers, Brandon; Papadatos, George; van Vlijmen, Herman W T; Kowalczyk, Wojtek; IJzerman, Adriaan P; van Westen, Gerard J P

2017-08-14

The increase of publicly available bioactivity data in recent years has fueled and catalyzed research in chemogenomics, data mining, and modeling approaches. As a direct result, over the past few years a multitude of different methods have been reported and evaluated, such as target fishing, nearest neighbor similarity-based methods, and Quantitative Structure Activity Relationship (QSAR)-based protocols. However, such studies are typically conducted on different datasets, using different validation strategies, and different metrics. In this study, different methods were compared using one single standardized dataset obtained from ChEMBL, which is made available to the public, using standardized metrics (BEDROC and Matthews Correlation Coefficient). Specifically, the performance of Naïve Bayes, Random Forests, Support Vector Machines, Logistic Regression, and Deep Neural Networks was assessed using QSAR and proteochemometric (PCM) methods. All methods were validated using both a random split validation and a temporal validation, with the latter being a more realistic benchmark of expected prospective execution. Deep Neural Networks are the top performing classifiers, highlighting the added value of Deep Neural Networks over other more conventional methods. Moreover, the best method ('DNN_PCM') performed significantly better at almost one standard deviation higher than the mean performance. Furthermore, Multi-task and PCM implementations were shown to improve performance over single task Deep Neural Networks. Conversely, target prediction performed almost two standard deviations under the mean performance. Random Forests, Support Vector Machines, and Logistic Regression performed around mean performance. Finally, using an ensemble of DNNs, alongside additional tuning, enhanced the relative performance by another 27% (compared with unoptimized 'DNN_PCM'). Here, a standardized set to test and evaluate different machine learning algorithms in the context of multi-task learning is offered by providing the data and the protocols. Graphical Abstract .

Deep learning classification in asteroseismology using an improved neural network: results on 15 000 Kepler red giants and applications to K2 and TESS data

NASA Astrophysics Data System (ADS)

Hon, Marc; Stello, Dennis; Yu, Jie

2018-05-01

Deep learning in the form of 1D convolutional neural networks have previously been shown to be capable of efficiently classifying the evolutionary state of oscillating red giants into red giant branch stars and helium-core burning stars by recognizing visual features in their asteroseismic frequency spectra. We elaborate further on the deep learning method by developing an improved convolutional neural network classifier. To make our method useful for current and future space missions such as K2, TESS, and PLATO, we train classifiers that are able to classify the evolutionary states of lower frequency resolution spectra expected from these missions. Additionally, we provide new classifications for 8633 Kepler red giants, out of which 426 have previously not been classified using asteroseismology. This brings the total to 14983 Kepler red giants classified with our new neural network. We also verify that our classifiers are remarkably robust to suboptimal data, including low signal-to-noise and incorrect training truth labels.

Automatic Seismic-Event Classification with Convolutional Neural Networks.

NASA Astrophysics Data System (ADS)

Bueno Rodriguez, A.; Titos Luzón, M.; Garcia Martinez, L.; Benitez, C.; Ibáñez, J. M.

2017-12-01

Active volcanoes exhibit a wide range of seismic signals, providing vast amounts of unlabelled volcano-seismic data that can be analyzed through the lens of artificial intelligence. However, obtaining high-quality labelled data is time-consuming and expensive. Deep neural networks can process data in their raw form, compute high-level features and provide a better representation of the input data distribution. These systems can be deployed to classify seismic data at scale, enhance current early-warning systems and build extensive seismic catalogs. In this research, we aim to classify spectrograms from seven different seismic events registered at "Volcán de Fuego" (Colima, Mexico), during four eruptive periods. Our approach is based on convolutional neural networks (CNNs), a sub-type of deep neural networks that can exploit grid structure from the data. Volcano-seismic signals can be mapped into a grid-like structure using the spectrogram: a representation of the temporal evolution in terms of time and frequency. Spectrograms were computed from the data using Hamming windows with 4 seconds length, 2.5 seconds overlapping and 128 points FFT resolution. Results are compared to deep neural networks, random forest and SVMs. Experiments show that CNNs can exploit temporal and frequency information, attaining a classification accuracy of 93%, similar to deep networks 91% but outperforming SVM and random forest. These results empirically show that CNNs are powerful models to classify a wide range of volcano-seismic signals, and achieve good generalization. Furthermore, volcano-seismic spectrograms contains useful discriminative information for the CNN, as higher layers of the network combine high-level features computed for each frequency band, helping to detect simultaneous events in time. Being at the intersection of deep learning and geophysics, this research enables future studies of how CNNs can be used in volcano monitoring to accurately determine the detection and location of seismic events.

Reduction of the dimension of neural network models in problems of pattern recognition and forecasting

NASA Astrophysics Data System (ADS)

Nasertdinova, A. D.; Bochkarev, V. V.

2017-11-01

Deep neural networks with a large number of parameters are a powerful tool for solving problems of pattern recognition, prediction and classification. Nevertheless, overfitting remains a serious problem in the use of such networks. A method of solving the problem of overfitting is proposed in this article. This method is based on reducing the number of independent parameters of a neural network model using the principal component analysis, and can be implemented using existing libraries of neural computing. The algorithm was tested on the problem of recognition of handwritten symbols from the MNIST database, as well as on the task of predicting time series (rows of the average monthly number of sunspots and series of the Lorentz system were used). It is shown that the application of the principal component analysis enables reducing the number of parameters of the neural network model when the results are good. The average error rate for the recognition of handwritten figures from the MNIST database was 1.12% (which is comparable to the results obtained using the "Deep training" methods), while the number of parameters of the neural network can be reduced to 130 times.

Dynamics of scene representations in the human brain revealed by magnetoencephalography and deep neural networks.

PubMed

Martin Cichy, Radoslaw; Khosla, Aditya; Pantazis, Dimitrios; Oliva, Aude

2017-06-01

Human scene recognition is a rapid multistep process evolving over time from single scene image to spatial layout processing. We used multivariate pattern analyses on magnetoencephalography (MEG) data to unravel the time course of this cortical process. Following an early signal for lower-level visual analysis of single scenes at ~100ms, we found a marker of real-world scene size, i.e. spatial layout processing, at ~250ms indexing neural representations robust to changes in unrelated scene properties and viewing conditions. For a quantitative model of how scene size representations may arise in the brain, we compared MEG data to a deep neural network model trained on scene classification. Representations of scene size emerged intrinsically in the model, and resolved emerging neural scene size representation. Together our data provide a first description of an electrophysiological signal for layout processing in humans, and suggest that deep neural networks are a promising framework to investigate how spatial layout representations emerge in the human brain. Copyright © 2017 The Authors. Published by Elsevier Inc. All rights reserved.

Deep Convolutional Neural Network-Based Early Automated Detection of Diabetic Retinopathy Using Fundus Image.

PubMed

Xu, Kele; Feng, Dawei; Mi, Haibo

2017-11-23

The automatic detection of diabetic retinopathy is of vital importance, as it is the main cause of irreversible vision loss in the working-age population in the developed world. The early detection of diabetic retinopathy occurrence can be very helpful for clinical treatment; although several different feature extraction approaches have been proposed, the classification task for retinal images is still tedious even for those trained clinicians. Recently, deep convolutional neural networks have manifested superior performance in image classification compared to previous handcrafted feature-based image classification methods. Thus, in this paper, we explored the use of deep convolutional neural network methodology for the automatic classification of diabetic retinopathy using color fundus image, and obtained an accuracy of 94.5% on our dataset, outperforming the results obtained by using classical approaches.

EMG-Based Estimation of Limb Movement Using Deep Learning With Recurrent Convolutional Neural Networks.

PubMed

Xia, Peng; Hu, Jie; Peng, Yinghong

2017-10-25

A novel model based on deep learning is proposed to estimate kinematic information for myoelectric control from multi-channel electromyogram (EMG) signals. The neural information of limb movement is embedded in EMG signals that are influenced by all kinds of factors. In order to overcome the negative effects of variability in signals, the proposed model employs the deep architecture combining convolutional neural networks (CNNs) and recurrent neural networks (RNNs). The EMG signals are transformed to time-frequency frames as the input to the model. The limb movement is estimated by the model that is trained with the gradient descent and backpropagation procedure. We tested the model for simultaneous and proportional estimation of limb movement in eight healthy subjects and compared it with support vector regression (SVR) and CNNs on the same data set. The experimental studies show that the proposed model has higher estimation accuracy and better robustness with respect to time. The combination of CNNs and RNNs can improve the model performance compared with using CNNs alone. The model of deep architecture is promising in EMG decoding and optimization of network structures can increase the accuracy and robustness. © 2017 International Center for Artificial Organs and Transplantation and Wiley Periodicals, Inc.

Background rejection in NEXT using deep neural networks

DOE PAGES

Renner, J.; Farbin, A.; Vidal, J. Muñoz; ...

2017-01-16

Here, we investigate the potential of using deep learning techniques to reject background events in searches for neutrinoless double beta decay with high pressure xenon time projection chambers capable of detailed track reconstruction. The differences in the topological signatures of background and signal events can be learned by deep neural networks via training over many thousands of events. These networks can then be used to classify further events as signal or background, providing an additional background rejection factor at an acceptable loss of efficiency. The networks trained in this study performed better than previous methods developed based on the usemore » of the same topological signatures by a factor of 1.2 to 1.6, and there is potential for further improvement.« less

Forecasting Flare Activity Using Deep Convolutional Neural Networks

NASA Astrophysics Data System (ADS)

Hernandez, T.

2017-12-01

Current operational flare forecasting relies on human morphological analysis of active regions and the persistence of solar flare activity through time (i.e. that the Sun will continue to do what it is doing right now: flaring or remaining calm). In this talk we present the results of applying deep Convolutional Neural Networks (CNNs) to the problem of solar flare forecasting. CNNs operate by training a set of tunable spatial filters that, in combination with neural layer interconnectivity, allow CNNs to automatically identify significant spatial structures predictive for classification and regression problems. We will start by discussing the applicability and success rate of the approach, the advantages it has over non-automated forecasts, and how mining our trained neural network provides a fresh look into the mechanisms behind magnetic energy storage and release.

Identification of Abnormal System Noise Temperature Patterns in Deep Space Network Antennas Using Neural Network Trained Fuzzy Logic

NASA Technical Reports Server (NTRS)

Lu, Thomas; Pham, Timothy; Liao, Jason

2011-01-01

This paper presents the development of a fuzzy logic function trained by an artificial neural network to classify the system noise temperature (SNT) of antennas in the NASA Deep Space Network (DSN). The SNT data were classified into normal, marginal, and abnormal classes. The irregular SNT pattern was further correlated with link margin and weather data. A reasonably good correlation is detected among high SNT, low link margin and the effect of bad weather; however we also saw some unexpected non-correlations which merit further study in the future.

Prediction of residue-residue contact matrix for protein-protein interaction with Fisher score features and deep learning.

PubMed

Du, Tianchuan; Liao, Li; Wu, Cathy H; Sun, Bilin

2016-11-01

Protein-protein interactions play essential roles in many biological processes. Acquiring knowledge of the residue-residue contact information of two interacting proteins is not only helpful in annotating functions for proteins, but also critical for structure-based drug design. The prediction of the protein residue-residue contact matrix of the interfacial regions is challenging. In this work, we introduced deep learning techniques (specifically, stacked autoencoders) to build deep neural network models to tackled the residue-residue contact prediction problem. In tandem with interaction profile Hidden Markov Models, which was used first to extract Fisher score features from protein sequences, stacked autoencoders were deployed to extract and learn hidden abstract features. The deep learning model showed significant improvement over the traditional machine learning model, Support Vector Machines (SVM), with the overall accuracy increased by 15% from 65.40% to 80.82%. We showed that the stacked autoencoders could extract novel features, which can be utilized by deep neural networks and other classifiers to enhance learning, out of the Fisher score features. It is further shown that deep neural networks have significant advantages over SVM in making use of the newly extracted features. Copyright © 2016. Published by Elsevier Inc.

Container-code recognition system based on computer vision and deep neural networks

NASA Astrophysics Data System (ADS)

Liu, Yi; Li, Tianjian; Jiang, Li; Liang, Xiaoyao

2018-04-01

Automatic container-code recognition system becomes a crucial requirement for ship transportation industry in recent years. In this paper, an automatic container-code recognition system based on computer vision and deep neural networks is proposed. The system consists of two modules, detection module and recognition module. The detection module applies both algorithms based on computer vision and neural networks, and generates a better detection result through combination to avoid the drawbacks of the two methods. The combined detection results are also collected for online training of the neural networks. The recognition module exploits both character segmentation and end-to-end recognition, and outputs the recognition result which passes the verification. When the recognition module generates false recognition, the result will be corrected and collected for online training of the end-to-end recognition sub-module. By combining several algorithms, the system is able to deal with more situations, and the online training mechanism can improve the performance of the neural networks at runtime. The proposed system is able to achieve 93% of overall recognition accuracy.

Detection of bars in galaxies using a deep convolutional neural network

NASA Astrophysics Data System (ADS)

Abraham, Sheelu; Aniyan, A. K.; Kembhavi, Ajit K.; Philip, N. S.; Vaghmare, Kaustubh

2018-06-01

We present an automated method for the detection of bar structure in optical images of galaxies using a deep convolutional neural network that is easy to use and provides good accuracy. In our study, we use a sample of 9346 galaxies in the redshift range of 0.009-0.2 from the Sloan Digital Sky Survey (SDSS), which has 3864 barred galaxies, the rest being unbarred. We reach a top precision of 94 per cent in identifying bars in galaxies using the trained network. This accuracy matches the accuracy reached by human experts on the same data without additional information about the images. Since deep convolutional neural networks can be scaled to handle large volumes of data, the method is expected to have great relevance in an era where astronomy data is rapidly increasing in terms of volume, variety, volatility, and velocity along with other V's that characterize big data. With the trained model, we have constructed a catalogue of barred galaxies from SDSS and made it available online.

Holography as deep learning

NASA Astrophysics Data System (ADS)

Gan, Wen-Cong; Shu, Fu-Wen

Quantum many-body problem with exponentially large degrees of freedom can be reduced to a tractable computational form by neural network method [G. Carleo and M. Troyer, Science 355 (2017) 602, arXiv:1606.02318.] The power of deep neural network (DNN) based on deep learning is clarified by mapping it to renormalization group (RG), which may shed lights on holographic principle by identifying a sequence of RG transformations to the AdS geometry. In this paper, we show that any network which reflects RG process has intrinsic hyperbolic geometry, and discuss the structure of entanglement encoded in the graph of DNN. We find the entanglement structure of DNN is of Ryu-Takayanagi form. Based on these facts, we argue that the emergence of holographic gravitational theory is related to deep learning process of the quantum-field theory.

Two Stage Data Augmentation for Low Resourced Speech Recognition (Author’s Manuscript)

DTIC Science & Technology

2016-09-12

speech recognition, deep neural networks, data augmentation 1. Introduction When training data is limited—whether it be audio or text—the obvious...Schwartz, and S. Tsakalidis, “Enhancing low resource keyword spotting with au- tomatically retrieved web documents,” in Interspeech, 2015, pp. 839–843. [2...and F. Seide, “Feature learning in deep neural networks - a study on speech recognition tasks,” in International Conference on Learning Representations

Boosting Contextual Information for Deep Neural Network Based Voice Activity Detection

DTIC Science & Technology

2015-02-01

multi-resolution stacking (MRS), which is a stack of ensemble classifiers. Each classifier in a building block inputs the concatenation of the predictions ...a base classifier in MRS, named boosted deep neural network (bDNN). bDNN first generates multiple base predictions from different contexts of a single...frame by only one DNN and then aggregates the base predictions for a better prediction of the frame, and it is different from computationally

Building on prior knowledge without building it in.

PubMed

Hansen, Steven S; Lampinen, Andrew K; Suri, Gaurav; McClelland, James L

2017-01-01

Lake et al. propose that people rely on "start-up software," "causal models," and "intuitive theories" built using compositional representations to learn new tasks more efficiently than some deep neural network models. We highlight the many drawbacks of a commitment to compositional representations and describe our continuing effort to explore how the ability to build on prior knowledge and to learn new tasks efficiently could arise through learning in deep neural networks.

Alexnet Feature Extraction and Multi-Kernel Learning for Objectoriented Classification

NASA Astrophysics Data System (ADS)

Ding, L.; Li, H.; Hu, C.; Zhang, W.; Wang, S.

2018-04-01

In view of the fact that the deep convolutional neural network has stronger ability of feature learning and feature expression, an exploratory research is done on feature extraction and classification for high resolution remote sensing images. Taking the Google image with 0.3 meter spatial resolution in Ludian area of Yunnan Province as an example, the image segmentation object was taken as the basic unit, and the pre-trained AlexNet deep convolution neural network model was used for feature extraction. And the spectral features, AlexNet features and GLCM texture features are combined with multi-kernel learning and SVM classifier, finally the classification results were compared and analyzed. The results show that the deep convolution neural network can extract more accurate remote sensing image features, and significantly improve the overall accuracy of classification, and provide a reference value for earthquake disaster investigation and remote sensing disaster evaluation.

Quicklook Constituent Abundance and Stretch Parameter Retrieval for the Juno Microwave Radiometer using Neural Networks

NASA Astrophysics Data System (ADS)

Bellotti, A.; Steffes, P. G.

2016-12-01

The Juno Microwave Radiometer (MWR) has six channels ranging from 1.36-50 cm and the ability to peer deep into the Jovian atmosphere. An Artifical Neural Network algorithm has been developed to rapidly perform inversion for the deep abundance of ammonia, the deep abundance of water vapor, and atmospheric "stretch" (a parameter that reflects the deviation from a wet adiabate in the higher atmosphere). This algorithm is "trained" by using simulated emissions at the six wavelengths computed using the Juno atmospheric microwave radiative transfer (JAMRT) model presented by Oyafuso et al. (This meeting). By exploiting the emission measurements conducted at six wavelengths and at various incident angles, the neural network can provide preliminary results to a useful precison in a computational method hundreds of times faster than conventional methods. This can quickly provide important insights into the variability and structure of the Jovian atmosphere.

Use of an Artificial Neural Network to Construct a Model of Predicting Deep Fungal Infection in Lung Cancer Patients.

PubMed

Chen, Jian; Chen, Jie; Ding, Hong-Yan; Pan, Qin-Shi; Hong, Wan-Dong; Xu, Gang; Yu, Fang-You; Wang, Yu-Min

2015-01-01

The statistical methods to analyze and predict the related dangerous factors of deep fungal infection in lung cancer patients were several, such as logic regression analysis, meta-analysis, multivariate Cox proportional hazards model analysis, retrospective analysis, and so on, but the results are inconsistent. A total of 696 patients with lung cancer were enrolled. The factors were compared employing Student's t-test or the Mann-Whitney test or the Chi-square test and variables that were significantly related to the presence of deep fungal infection selected as candidates for input into the final artificial neural network analysis (ANN) model. The receiver operating characteristic (ROC) and area under curve (AUC) were used to evaluate the performance of the artificial neural network (ANN) model and logistic regression (LR) model. The prevalence of deep fungal infection from lung cancer in this entire study population was 32.04%(223/696), deep fungal infections occur in sputum specimens 44.05% (200/454). The ratio of candida albicans was 86.99% (194/223) in the total fungi. It was demonstrated that older (≥65 years), use of antibiotics, low serum albumin concentrations (≤37.18 g /L), radiotherapy, surgery, low hemoglobin hyperlipidemia (≤93.67 g /L), long time of hospitalization (≥14 days) were apt to deep fungal infection and the ANN model consisted of the seven factors. The AUC of ANN model (0.829±0.019) was higher than that of LR model (0.756±0.021). The artificial neural network model with variables consisting of age, use of antibiotics, serum albumin concentrations, received radiotherapy, received surgery, hemoglobin, time of hospitalization should be useful for predicting the deep fungal infection in lung cancer.

Deep Potential Molecular Dynamics: A Scalable Model with the Accuracy of Quantum Mechanics

NASA Astrophysics Data System (ADS)

Zhang, Linfeng; Han, Jiequn; Wang, Han; Car, Roberto; E, Weinan

2018-04-01

We introduce a scheme for molecular simulations, the deep potential molecular dynamics (DPMD) method, based on a many-body potential and interatomic forces generated by a carefully crafted deep neural network trained with ab initio data. The neural network model preserves all the natural symmetries in the problem. It is first-principles based in the sense that there are no ad hoc components aside from the network model. We show that the proposed scheme provides an efficient and accurate protocol in a variety of systems, including bulk materials and molecules. In all these cases, DPMD gives results that are essentially indistinguishable from the original data, at a cost that scales linearly with system size.

Scalable training of artificial neural networks with adaptive sparse connectivity inspired by network science.

PubMed

Mocanu, Decebal Constantin; Mocanu, Elena; Stone, Peter; Nguyen, Phuong H; Gibescu, Madeleine; Liotta, Antonio

2018-06-19

Through the success of deep learning in various domains, artificial neural networks are currently among the most used artificial intelligence methods. Taking inspiration from the network properties of biological neural networks (e.g. sparsity, scale-freeness), we argue that (contrary to general practice) artificial neural networks, too, should not have fully-connected layers. Here we propose sparse evolutionary training of artificial neural networks, an algorithm which evolves an initial sparse topology (Erdős-Rényi random graph) of two consecutive layers of neurons into a scale-free topology, during learning. Our method replaces artificial neural networks fully-connected layers with sparse ones before training, reducing quadratically the number of parameters, with no decrease in accuracy. We demonstrate our claims on restricted Boltzmann machines, multi-layer perceptrons, and convolutional neural networks for unsupervised and supervised learning on 15 datasets. Our approach has the potential to enable artificial neural networks to scale up beyond what is currently possible.

OCT-based deep learning algorithm for the evaluation of treatment indication with anti-vascular endothelial growth factor medications.

PubMed

Prahs, Philipp; Radeck, Viola; Mayer, Christian; Cvetkov, Yordan; Cvetkova, Nadezhda; Helbig, Horst; Märker, David

2018-01-01

Intravitreal injections with anti-vascular endothelial growth factor (anti-VEGF) medications have become the standard of care for their respective indications. Optical coherence tomography (OCT) scans of the central retina provide detailed anatomical data and are widely used by clinicians in the decision-making process of anti-VEGF indication. In recent years, significant progress has been made in artificial intelligence and computer vision research. We trained a deep convolutional artificial neural network to predict treatment indication based on central retinal OCT scans without human intervention. A total of 183,402 retinal OCT B-scans acquired between 2008 and 2016 were exported from the institutional image archive of a university hospital. OCT images were cross-referenced with the electronic institutional intravitreal injection records. OCT images with a following intravitreal injection during the first 21 days after image acquisition were assigned into the 'injection' group, while the same amount of random OCT images without intravitreal injections was labeled as 'no injection'. After image preprocessing, OCT images were split in a 9:1 ratio to training and test datasets. We trained a GoogLeNet inception deep convolutional neural network and assessed its performance on the validation dataset. We calculated prediction accuracy, sensitivity, specificity, and receiver operating characteristics. The deep convolutional neural network was successfully trained on the extracted clinical data. The trained neural network classifier reached a prediction accuracy of 95.5% on the images in the validation dataset. For single retinal B-scans in the validation dataset, a sensitivity of 90.1% and a specificity of 96.2% were achieved. The area under the receiver operating characteristic curve was 0.968 on a per B-scan image basis, and 0.988 by averaging over six B-scans per examination on the validation dataset. Deep artificial neural networks show impressive performance on classification of retinal OCT scans. After training on historical clinical data, machine learning methods can offer the clinician support in the decision-making process. Care should be taken not to mistake neural network output as treatment recommendation and to ensure a final thorough evaluation by the treating physician.

Assessing the Linguistic Productivity of Unsupervised Deep Neural Networks

DOE Office of Scientific and Technical Information (OSTI.GOV)

Phillips, Lawrence A.; Hodas, Nathan O.

Increasingly, cognitive scientists have demonstrated interest in applying tools from deep learning. One use for deep learning is in language acquisition where it is useful to know if a linguistic phenomenon can be learned through domain-general means. To assess whether unsupervised deep learning is appropriate, we first pose a smaller question: Can unsupervised neural networks apply linguistic rules productively, using them in novel situations. We draw from the literature on determiner/noun productivity by training an unsupervised, autoencoder network measuring its ability to combine nouns with determiners. Our simple autoencoder creates combinations it has not previously encountered, displaying a degree ofmore » overlap similar to actual children. While this preliminary work does not provide conclusive evidence for productivity, it warrants further investigation with more complex models. Further, this work helps lay the foundations for future collaboration between the deep learning and cognitive science communities.« less

Deep Recurrent Neural Networks for Human Activity Recognition

PubMed Central

Murad, Abdulmajid

2017-01-01

Adopting deep learning methods for human activity recognition has been effective in extracting discriminative features from raw input sequences acquired from body-worn sensors. Although human movements are encoded in a sequence of successive samples in time, typical machine learning methods perform recognition tasks without exploiting the temporal correlations between input data samples. Convolutional neural networks (CNNs) address this issue by using convolutions across a one-dimensional temporal sequence to capture dependencies among input data. However, the size of convolutional kernels restricts the captured range of dependencies between data samples. As a result, typical models are unadaptable to a wide range of activity-recognition configurations and require fixed-length input windows. In this paper, we propose the use of deep recurrent neural networks (DRNNs) for building recognition models that are capable of capturing long-range dependencies in variable-length input sequences. We present unidirectional, bidirectional, and cascaded architectures based on long short-term memory (LSTM) DRNNs and evaluate their effectiveness on miscellaneous benchmark datasets. Experimental results show that our proposed models outperform methods employing conventional machine learning, such as support vector machine (SVM) and k-nearest neighbors (KNN). Additionally, the proposed models yield better performance than other deep learning techniques, such as deep believe networks (DBNs) and CNNs. PMID:29113103

Deep Recurrent Neural Networks for Human Activity Recognition.

PubMed

Murad, Abdulmajid; Pyun, Jae-Young

2017-11-06

Adopting deep learning methods for human activity recognition has been effective in extracting discriminative features from raw input sequences acquired from body-worn sensors. Although human movements are encoded in a sequence of successive samples in time, typical machine learning methods perform recognition tasks without exploiting the temporal correlations between input data samples. Convolutional neural networks (CNNs) address this issue by using convolutions across a one-dimensional temporal sequence to capture dependencies among input data. However, the size of convolutional kernels restricts the captured range of dependencies between data samples. As a result, typical models are unadaptable to a wide range of activity-recognition configurations and require fixed-length input windows. In this paper, we propose the use of deep recurrent neural networks (DRNNs) for building recognition models that are capable of capturing long-range dependencies in variable-length input sequences. We present unidirectional, bidirectional, and cascaded architectures based on long short-term memory (LSTM) DRNNs and evaluate their effectiveness on miscellaneous benchmark datasets. Experimental results show that our proposed models outperform methods employing conventional machine learning, such as support vector machine (SVM) and k-nearest neighbors (KNN). Additionally, the proposed models yield better performance than other deep learning techniques, such as deep believe networks (DBNs) and CNNs.

LiteNet: Lightweight Neural Network for Detecting Arrhythmias at Resource-Constrained Mobile Devices.

PubMed

He, Ziyang; Zhang, Xiaoqing; Cao, Yangjie; Liu, Zhi; Zhang, Bo; Wang, Xiaoyan

2018-04-17

By running applications and services closer to the user, edge processing provides many advantages, such as short response time and reduced network traffic. Deep-learning based algorithms provide significantly better performances than traditional algorithms in many fields but demand more resources, such as higher computational power and more memory. Hence, designing deep learning algorithms that are more suitable for resource-constrained mobile devices is vital. In this paper, we build a lightweight neural network, termed LiteNet which uses a deep learning algorithm design to diagnose arrhythmias, as an example to show how we design deep learning schemes for resource-constrained mobile devices. Compare to other deep learning models with an equivalent accuracy, LiteNet has several advantages. It requires less memory, incurs lower computational cost, and is more feasible for deployment on resource-constrained mobile devices. It can be trained faster than other neural network algorithms and requires less communication across different processing units during distributed training. It uses filters of heterogeneous size in a convolutional layer, which contributes to the generation of various feature maps. The algorithm was tested using the MIT-BIH electrocardiogram (ECG) arrhythmia database; the results showed that LiteNet outperforms comparable schemes in diagnosing arrhythmias, and in its feasibility for use at the mobile devices.

LiteNet: Lightweight Neural Network for Detecting Arrhythmias at Resource-Constrained Mobile Devices

PubMed Central

Zhang, Xiaoqing; Cao, Yangjie; Liu, Zhi; Zhang, Bo; Wang, Xiaoyan

2018-01-01

By running applications and services closer to the user, edge processing provides many advantages, such as short response time and reduced network traffic. Deep-learning based algorithms provide significantly better performances than traditional algorithms in many fields but demand more resources, such as higher computational power and more memory. Hence, designing deep learning algorithms that are more suitable for resource-constrained mobile devices is vital. In this paper, we build a lightweight neural network, termed LiteNet which uses a deep learning algorithm design to diagnose arrhythmias, as an example to show how we design deep learning schemes for resource-constrained mobile devices. Compare to other deep learning models with an equivalent accuracy, LiteNet has several advantages. It requires less memory, incurs lower computational cost, and is more feasible for deployment on resource-constrained mobile devices. It can be trained faster than other neural network algorithms and requires less communication across different processing units during distributed training. It uses filters of heterogeneous size in a convolutional layer, which contributes to the generation of various feature maps. The algorithm was tested using the MIT-BIH electrocardiogram (ECG) arrhythmia database; the results showed that LiteNet outperforms comparable schemes in diagnosing arrhythmias, and in its feasibility for use at the mobile devices. PMID:29673171

An introduction to deep learning on biological sequence data: examples and solutions.

PubMed

Jurtz, Vanessa Isabell; Johansen, Alexander Rosenberg; Nielsen, Morten; Almagro Armenteros, Jose Juan; Nielsen, Henrik; Sønderby, Casper Kaae; Winther, Ole; Sønderby, Søren Kaae

2017-11-15

Deep neural network architectures such as convolutional and long short-term memory networks have become increasingly popular as machine learning tools during the recent years. The availability of greater computational resources, more data, new algorithms for training deep models and easy to use libraries for implementation and training of neural networks are the drivers of this development. The use of deep learning has been especially successful in image recognition; and the development of tools, applications and code examples are in most cases centered within this field rather than within biology. Here, we aim to further the development of deep learning methods within biology by providing application examples and ready to apply and adapt code templates. Given such examples, we illustrate how architectures consisting of convolutional and long short-term memory neural networks can relatively easily be designed and trained to state-of-the-art performance on three biological sequence problems: prediction of subcellular localization, protein secondary structure and the binding of peptides to MHC Class II molecules. All implementations and datasets are available online to the scientific community at https://github.com/vanessajurtz/lasagne4bio. skaaesonderby@gmail.com. Supplementary data are available at Bioinformatics online. © The Author (2017). Published by Oxford University Press. All rights reserved. For Permissions, please email: journals.permissions@oup.com

Do deep convolutional neural networks really need to be deep when applied for remote scene classification?

NASA Astrophysics Data System (ADS)

Luo, Chang; Wang, Jie; Feng, Gang; Xu, Suhui; Wang, Shiqiang

2017-10-01

Deep convolutional neural networks (CNNs) have been widely used to obtain high-level representation in various computer vision tasks. However, for remote scene classification, there are not sufficient images to train a very deep CNN from scratch. From two viewpoints of generalization power, we propose two promising kinds of deep CNNs for remote scenes and try to find whether deep CNNs need to be deep for remote scene classification. First, we transfer successful pretrained deep CNNs to remote scenes based on the theory that depth of CNNs brings the generalization power by learning available hypothesis for finite data samples. Second, according to the opposite viewpoint that generalization power of deep CNNs comes from massive memorization and shallow CNNs with enough neural nodes have perfect finite sample expressivity, we design a lightweight deep CNN (LDCNN) for remote scene classification. With five well-known pretrained deep CNNs, experimental results on two independent remote-sensing datasets demonstrate that transferred deep CNNs can achieve state-of-the-art results in an unsupervised setting. However, because of its shallow architecture, LDCNN cannot obtain satisfactory performance, regardless of whether in an unsupervised, semisupervised, or supervised setting. CNNs really need depth to obtain general features for remote scenes. This paper also provides baseline for applying deep CNNs to other remote sensing tasks.

Shallow Transits—Deep Learning. I. Feasibility Study of Deep Learning to Detect Periodic Transits of Exoplanets

NASA Astrophysics Data System (ADS)

Zucker, Shay; Giryes, Raja

2018-04-01

Transits of habitable planets around solar-like stars are expected to be shallow, and to have long periods, which means low information content. The current bottleneck in the detection of such transits is caused in large part by the presence of red (correlated) noise in the light curves obtained from the dedicated space telescopes. Based on the groundbreaking results deep learning achieves in many signal and image processing applications, we propose to use deep neural networks to solve this problem. We present a feasibility study, in which we applied a convolutional neural network on a simulated training set. The training set comprised light curves received from a hypothetical high-cadence space-based telescope. We simulated the red noise by using Gaussian Processes with a wide variety of hyper-parameters. We then tested the network on a completely different test set simulated in the same way. Our study proves that very difficult cases can indeed be detected. Furthermore, we show how detection trends can be studied and detection biases quantified. We have also checked the robustness of the neural-network performance against practical artifacts such as outliers and discontinuities, which are known to affect space-based high-cadence light curves. Future work will allow us to use the neural networks to characterize the transit model and identify individual transits. This new approach will certainly be an indispensable tool for the detection of habitable planets in the future planet-detection space missions such as PLATO.

Pubface: Celebrity face identification based on deep learning

NASA Astrophysics Data System (ADS)

Ouanan, H.; Ouanan, M.; Aksasse, B.

2018-05-01

In this paper, we describe a new real time application called PubFace, which allows to recognize celebrities in public spaces by employs a new pose invariant face recognition deep neural network algorithm with an extremely low error rate. To build this application, we make the following contributions: firstly, we build a novel dataset with over five million faces labelled. Secondly, we fine tuning the deep convolutional neural network (CNN) VGG-16 architecture on our new dataset that we have built. Finally, we deploy this model on the Raspberry Pi 3 model B using the OpenCv dnn module (OpenCV 3.3).

A Spatiotemporal Prediction Framework for Air Pollution Based on Deep RNN

NASA Astrophysics Data System (ADS)

Fan, J.; Li, Q.; Hou, J.; Feng, X.; Karimian, H.; Lin, S.

2017-10-01

Time series data in practical applications always contain missing values due to sensor malfunction, network failure, outliers etc. In order to handle missing values in time series, as well as the lack of considering temporal properties in machine learning models, we propose a spatiotemporal prediction framework based on missing value processing algorithms and deep recurrent neural network (DRNN). By using missing tag and missing interval to represent time series patterns, we implement three different missing value fixing algorithms, which are further incorporated into deep neural network that consists of LSTM (Long Short-term Memory) layers and fully connected layers. Real-world air quality and meteorological datasets (Jingjinji area, China) are used for model training and testing. Deep feed forward neural networks (DFNN) and gradient boosting decision trees (GBDT) are trained as baseline models against the proposed DRNN. Performances of three missing value fixing algorithms, as well as different machine learning models are evaluated and analysed. Experiments show that the proposed DRNN framework outperforms both DFNN and GBDT, therefore validating the capacity of the proposed framework. Our results also provides useful insights for better understanding of different strategies that handle missing values.

Deep Convolutional and LSTM Recurrent Neural Networks for Multimodal Wearable Activity Recognition.

PubMed

Ordóñez, Francisco Javier; Roggen, Daniel

2016-01-18

Human activity recognition (HAR) tasks have traditionally been solved using engineered features obtained by heuristic processes. Current research suggests that deep convolutional neural networks are suited to automate feature extraction from raw sensor inputs. However, human activities are made of complex sequences of motor movements, and capturing this temporal dynamics is fundamental for successful HAR. Based on the recent success of recurrent neural networks for time series domains, we propose a generic deep framework for activity recognition based on convolutional and LSTM recurrent units, which: (i) is suitable for multimodal wearable sensors; (ii) can perform sensor fusion naturally; (iii) does not require expert knowledge in designing features; and (iv) explicitly models the temporal dynamics of feature activations. We evaluate our framework on two datasets, one of which has been used in a public activity recognition challenge. Our results show that our framework outperforms competing deep non-recurrent networks on the challenge dataset by 4% on average; outperforming some of the previous reported results by up to 9%. Our results show that the framework can be applied to homogeneous sensor modalities, but can also fuse multimodal sensors to improve performance. We characterise key architectural hyperparameters' influence on performance to provide insights about their optimisation.

Speech reconstruction using a deep partially supervised neural network.

PubMed

McLoughlin, Ian; Li, Jingjie; Song, Yan; Sharifzadeh, Hamid R

2017-08-01

Statistical speech reconstruction for larynx-related dysphonia has achieved good performance using Gaussian mixture models and, more recently, restricted Boltzmann machine arrays; however, deep neural network (DNN)-based systems have been hampered by the limited amount of training data available from individual voice-loss patients. The authors propose a novel DNN structure that allows a partially supervised training approach on spectral features from smaller data sets, yielding very good results compared with the current state-of-the-art.

Deep convolutional neural network for prostate MR segmentation

NASA Astrophysics Data System (ADS)

Tian, Zhiqiang; Liu, Lizhi; Fei, Baowei

2017-03-01

Automatic segmentation of the prostate in magnetic resonance imaging (MRI) has many applications in prostate cancer diagnosis and therapy. We propose a deep fully convolutional neural network (CNN) to segment the prostate automatically. Our deep CNN model is trained end-to-end in a single learning stage based on prostate MR images and the corresponding ground truths, and learns to make inference for pixel-wise segmentation. Experiments were performed on our in-house data set, which contains prostate MR images of 20 patients. The proposed CNN model obtained a mean Dice similarity coefficient of 85.3%+/-3.2% as compared to the manual segmentation. Experimental results show that our deep CNN model could yield satisfactory segmentation of the prostate.

Do neural nets learn statistical laws behind natural language?

PubMed

Takahashi, Shuntaro; Tanaka-Ishii, Kumiko

2017-01-01

The performance of deep learning in natural language processing has been spectacular, but the reasons for this success remain unclear because of the inherent complexity of deep learning. This paper provides empirical evidence of its effectiveness and of a limitation of neural networks for language engineering. Precisely, we demonstrate that a neural language model based on long short-term memory (LSTM) effectively reproduces Zipf's law and Heaps' law, two representative statistical properties underlying natural language. We discuss the quality of reproducibility and the emergence of Zipf's law and Heaps' law as training progresses. We also point out that the neural language model has a limitation in reproducing long-range correlation, another statistical property of natural language. This understanding could provide a direction for improving the architectures of neural networks.

Do neural nets learn statistical laws behind natural language?

PubMed Central

Takahashi, Shuntaro

2017-01-01

The performance of deep learning in natural language processing has been spectacular, but the reasons for this success remain unclear because of the inherent complexity of deep learning. This paper provides empirical evidence of its effectiveness and of a limitation of neural networks for language engineering. Precisely, we demonstrate that a neural language model based on long short-term memory (LSTM) effectively reproduces Zipf’s law and Heaps’ law, two representative statistical properties underlying natural language. We discuss the quality of reproducibility and the emergence of Zipf’s law and Heaps’ law as training progresses. We also point out that the neural language model has a limitation in reproducing long-range correlation, another statistical property of natural language. This understanding could provide a direction for improving the architectures of neural networks. PMID:29287076

Deep RNNs for video denoising

NASA Astrophysics Data System (ADS)

Chen, Xinyuan; Song, Li; Yang, Xiaokang

2016-09-01

Video denoising can be described as the problem of mapping from a specific length of noisy frames to clean one. We propose a deep architecture based on Recurrent Neural Network (RNN) for video denoising. The model learns a patch-based end-to-end mapping between the clean and noisy video sequences. It takes the corrupted video sequences as the input and outputs the clean one. Our deep network, which we refer to as deep Recurrent Neural Networks (deep RNNs or DRNNs), stacks RNN layers where each layer receives the hidden state of the previous layer as input. Experiment shows (i) the recurrent architecture through temporal domain extracts motion information and does favor to video denoising, and (ii) deep architecture have large enough capacity for expressing mapping relation between corrupted videos as input and clean videos as output, furthermore, (iii) the model has generality to learned different mappings from videos corrupted by different types of noise (e.g., Poisson-Gaussian noise). By training on large video databases, we are able to compete with some existing video denoising methods.

Searching for exoplanets using artificial intelligence

NASA Astrophysics Data System (ADS)

Pearson, Kyle A.; Palafox, Leon; Griffith, Caitlin A.

2018-02-01

In the last decade, over a million stars were monitored to detect transiting planets. Manual interpretation of potential exoplanet candidates is labor intensive and subject to human error, the results of which are difficult to quantify. Here we present a new method of detecting exoplanet candidates in large planetary search projects which, unlike current methods uses a neural network. Neural networks, also called "deep learning" or "deep nets" are designed to give a computer perception into a specific problem by training it to recognize patterns. Unlike past transit detection algorithms deep nets learn to recognize planet features instead of relying on hand-coded metrics that humans perceive as the most representative. Our convolutional neural network is capable of detecting Earth-like exoplanets in noisy time-series data with a greater accuracy than a least-squares method. Deep nets are highly generalizable allowing data to be evaluated from different time series after interpolation without compromising performance. As validated by our deep net analysis of Kepler light curves, we detect periodic transits consistent with the true period without any model fitting. Our study indicates that machine learning will facilitate the characterization of exoplanets in future analysis of large astronomy data sets.

Multiview fusion for activity recognition using deep neural networks

NASA Astrophysics Data System (ADS)

Kavi, Rahul; Kulathumani, Vinod; Rohit, Fnu; Kecojevic, Vlad

2016-07-01

Convolutional neural networks (ConvNets) coupled with long short term memory (LSTM) networks have been recently shown to be effective for video classification as they combine the automatic feature extraction capabilities of a neural network with additional memory in the temporal domain. This paper shows how multiview fusion can be applied to such a ConvNet LSTM architecture. Two different fusion techniques are presented. The system is first evaluated in the context of a driver activity recognition system using data collected in a multicamera driving simulator. These results show significant improvement in accuracy with multiview fusion and also show that deep learning performs better than a traditional approach using spatiotemporal features even without requiring any background subtraction. The system is also validated on another publicly available multiview action recognition dataset that has 12 action classes and 8 camera views.

Single-shot T2 mapping using overlapping-echo detachment planar imaging and a deep convolutional neural network.

PubMed

Cai, Congbo; Wang, Chao; Zeng, Yiqing; Cai, Shuhui; Liang, Dong; Wu, Yawen; Chen, Zhong; Ding, Xinghao; Zhong, Jianhui

2018-04-24

An end-to-end deep convolutional neural network (CNN) based on deep residual network (ResNet) was proposed to efficiently reconstruct reliable T 2 mapping from single-shot overlapping-echo detachment (OLED) planar imaging. The training dataset was obtained from simulations that were carried out on SPROM (Simulation with PRoduct Operator Matrix) software developed by our group. The relationship between the original OLED image containing two echo signals and the corresponding T 2 mapping was learned by ResNet training. After the ResNet was trained, it was applied to reconstruct the T 2 mapping from simulation and in vivo human brain data. Although the ResNet was trained entirely on simulated data, the trained network was generalized well to real human brain data. The results from simulation and in vivo human brain experiments show that the proposed method significantly outperforms the echo-detachment-based method. Reliable T 2 mapping with higher accuracy is achieved within 30 ms after the network has been trained, while the echo-detachment-based OLED reconstruction method took approximately 2 min. The proposed method will facilitate real-time dynamic and quantitative MR imaging via OLED sequence, and deep convolutional neural network has the potential to reconstruct maps from complex MRI sequences efficiently. © 2018 International Society for Magnetic Resonance in Medicine.

Event-Driven Random Back-Propagation: Enabling Neuromorphic Deep Learning Machines

PubMed Central

Neftci, Emre O.; Augustine, Charles; Paul, Somnath; Detorakis, Georgios

2017-01-01

An ongoing challenge in neuromorphic computing is to devise general and computationally efficient models of inference and learning which are compatible with the spatial and temporal constraints of the brain. One increasingly popular and successful approach is to take inspiration from inference and learning algorithms used in deep neural networks. However, the workhorse of deep learning, the gradient descent Gradient Back Propagation (BP) rule, often relies on the immediate availability of network-wide information stored with high-precision memory during learning, and precise operations that are difficult to realize in neuromorphic hardware. Remarkably, recent work showed that exact backpropagated gradients are not essential for learning deep representations. Building on these results, we demonstrate an event-driven random BP (eRBP) rule that uses an error-modulated synaptic plasticity for learning deep representations. Using a two-compartment Leaky Integrate & Fire (I&F) neuron, the rule requires only one addition and two comparisons for each synaptic weight, making it very suitable for implementation in digital or mixed-signal neuromorphic hardware. Our results show that using eRBP, deep representations are rapidly learned, achieving classification accuracies on permutation invariant datasets comparable to those obtained in artificial neural network simulations on GPUs, while being robust to neural and synaptic state quantizations during learning. PMID:28680387

Event-Driven Random Back-Propagation: Enabling Neuromorphic Deep Learning Machines.

PubMed

Neftci, Emre O; Augustine, Charles; Paul, Somnath; Detorakis, Georgios

2017-01-01

An ongoing challenge in neuromorphic computing is to devise general and computationally efficient models of inference and learning which are compatible with the spatial and temporal constraints of the brain. One increasingly popular and successful approach is to take inspiration from inference and learning algorithms used in deep neural networks. However, the workhorse of deep learning, the gradient descent Gradient Back Propagation (BP) rule, often relies on the immediate availability of network-wide information stored with high-precision memory during learning, and precise operations that are difficult to realize in neuromorphic hardware. Remarkably, recent work showed that exact backpropagated gradients are not essential for learning deep representations. Building on these results, we demonstrate an event-driven random BP (eRBP) rule that uses an error-modulated synaptic plasticity for learning deep representations. Using a two-compartment Leaky Integrate & Fire (I&F) neuron, the rule requires only one addition and two comparisons for each synaptic weight, making it very suitable for implementation in digital or mixed-signal neuromorphic hardware. Our results show that using eRBP, deep representations are rapidly learned, achieving classification accuracies on permutation invariant datasets comparable to those obtained in artificial neural network simulations on GPUs, while being robust to neural and synaptic state quantizations during learning.

Deep neural models for ICD-10 coding of death certificates and autopsy reports in free-text.

PubMed

Duarte, Francisco; Martins, Bruno; Pinto, Cátia Sousa; Silva, Mário J

2018-04-01

We address the assignment of ICD-10 codes for causes of death by analyzing free-text descriptions in death certificates, together with the associated autopsy reports and clinical bulletins, from the Portuguese Ministry of Health. We leverage a deep neural network that combines word embeddings, recurrent units, and neural attention, for the generation of intermediate representations of the textual contents. The neural network also explores the hierarchical nature of the input data, by building representations from the sequences of words within individual fields, which are then combined according to the sequences of fields that compose the inputs. Moreover, we explore innovative mechanisms for initializing the weights of the final nodes of the network, leveraging co-occurrences between classes together with the hierarchical structure of ICD-10. Experimental results attest to the contribution of the different neural network components. Our best model achieves accuracy scores over 89%, 81%, and 76%, respectively for ICD-10 chapters, blocks, and full-codes. Through examples, we also show that our method can produce interpretable results, useful for public health surveillance. Copyright © 2018 Elsevier Inc. All rights reserved.

Protein Secondary Structure Prediction Using Deep Convolutional Neural Fields.

PubMed

Wang, Sheng; Peng, Jian; Ma, Jianzhu; Xu, Jinbo

2016-01-11

Protein secondary structure (SS) prediction is important for studying protein structure and function. When only the sequence (profile) information is used as input feature, currently the best predictors can obtain ~80% Q3 accuracy, which has not been improved in the past decade. Here we present DeepCNF (Deep Convolutional Neural Fields) for protein SS prediction. DeepCNF is a Deep Learning extension of Conditional Neural Fields (CNF), which is an integration of Conditional Random Fields (CRF) and shallow neural networks. DeepCNF can model not only complex sequence-structure relationship by a deep hierarchical architecture, but also interdependency between adjacent SS labels, so it is much more powerful than CNF. Experimental results show that DeepCNF can obtain ~84% Q3 accuracy, ~85% SOV score, and ~72% Q8 accuracy, respectively, on the CASP and CAMEO test proteins, greatly outperforming currently popular predictors. As a general framework, DeepCNF can be used to predict other protein structure properties such as contact number, disorder regions, and solvent accessibility.

Protein Secondary Structure Prediction Using Deep Convolutional Neural Fields

NASA Astrophysics Data System (ADS)

Wang, Sheng; Peng, Jian; Ma, Jianzhu; Xu, Jinbo

2016-01-01

Protein secondary structure (SS) prediction is important for studying protein structure and function. When only the sequence (profile) information is used as input feature, currently the best predictors can obtain ~80% Q3 accuracy, which has not been improved in the past decade. Here we present DeepCNF (Deep Convolutional Neural Fields) for protein SS prediction. DeepCNF is a Deep Learning extension of Conditional Neural Fields (CNF), which is an integration of Conditional Random Fields (CRF) and shallow neural networks. DeepCNF can model not only complex sequence-structure relationship by a deep hierarchical architecture, but also interdependency between adjacent SS labels, so it is much more powerful than CNF. Experimental results show that DeepCNF can obtain ~84% Q3 accuracy, ~85% SOV score, and ~72% Q8 accuracy, respectively, on the CASP and CAMEO test proteins, greatly outperforming currently popular predictors. As a general framework, DeepCNF can be used to predict other protein structure properties such as contact number, disorder regions, and solvent accessibility.

Modeling positional effects of regulatory sequences with spline transformations increases prediction accuracy of deep neural networks

PubMed Central

Avsec, Žiga; Cheng, Jun; Gagneur, Julien

2018-01-01

Abstract Motivation Regulatory sequences are not solely defined by their nucleic acid sequence but also by their relative distances to genomic landmarks such as transcription start site, exon boundaries or polyadenylation site. Deep learning has become the approach of choice for modeling regulatory sequences because of its strength to learn complex sequence features. However, modeling relative distances to genomic landmarks in deep neural networks has not been addressed. Results Here we developed spline transformation, a neural network module based on splines to flexibly and robustly model distances. Modeling distances to various genomic landmarks with spline transformations significantly increased state-of-the-art prediction accuracy of in vivo RNA-binding protein binding sites for 120 out of 123 proteins. We also developed a deep neural network for human splice branchpoint based on spline transformations that outperformed the current best, already distance-based, machine learning model. Compared to piecewise linear transformation, as obtained by composition of rectified linear units, spline transformation yields higher prediction accuracy as well as faster and more robust training. As spline transformation can be applied to further quantities beyond distances, such as methylation or conservation, we foresee it as a versatile component in the genomics deep learning toolbox. Availability and implementation Spline transformation is implemented as a Keras layer in the CONCISE python package: https://github.com/gagneurlab/concise. Analysis code is available at https://github.com/gagneurlab/Manuscript_Avsec_Bioinformatics_2017. Contact avsec@in.tum.de or gagneur@in.tum.de Supplementary information Supplementary data are available at Bioinformatics online. PMID:29155928

Conversion of Continuous-Valued Deep Networks to Efficient Event-Driven Networks for Image Classification.

PubMed

Rueckauer, Bodo; Lungu, Iulia-Alexandra; Hu, Yuhuang; Pfeiffer, Michael; Liu, Shih-Chii

2017-01-01

Spiking neural networks (SNNs) can potentially offer an efficient way of doing inference because the neurons in the networks are sparsely activated and computations are event-driven. Previous work showed that simple continuous-valued deep Convolutional Neural Networks (CNNs) can be converted into accurate spiking equivalents. These networks did not include certain common operations such as max-pooling, softmax, batch-normalization and Inception-modules. This paper presents spiking equivalents of these operations therefore allowing conversion of nearly arbitrary CNN architectures. We show conversion of popular CNN architectures, including VGG-16 and Inception-v3, into SNNs that produce the best results reported to date on MNIST, CIFAR-10 and the challenging ImageNet dataset. SNNs can trade off classification error rate against the number of available operations whereas deep continuous-valued neural networks require a fixed number of operations to achieve their classification error rate. From the examples of LeNet for MNIST and BinaryNet for CIFAR-10, we show that with an increase in error rate of a few percentage points, the SNNs can achieve more than 2x reductions in operations compared to the original CNNs. This highlights the potential of SNNs in particular when deployed on power-efficient neuromorphic spiking neuron chips, for use in embedded applications.

Conversion of Continuous-Valued Deep Networks to Efficient Event-Driven Networks for Image Classification

PubMed Central

Rueckauer, Bodo; Lungu, Iulia-Alexandra; Hu, Yuhuang; Pfeiffer, Michael; Liu, Shih-Chii

2017-01-01

Spiking neural networks (SNNs) can potentially offer an efficient way of doing inference because the neurons in the networks are sparsely activated and computations are event-driven. Previous work showed that simple continuous-valued deep Convolutional Neural Networks (CNNs) can be converted into accurate spiking equivalents. These networks did not include certain common operations such as max-pooling, softmax, batch-normalization and Inception-modules. This paper presents spiking equivalents of these operations therefore allowing conversion of nearly arbitrary CNN architectures. We show conversion of popular CNN architectures, including VGG-16 and Inception-v3, into SNNs that produce the best results reported to date on MNIST, CIFAR-10 and the challenging ImageNet dataset. SNNs can trade off classification error rate against the number of available operations whereas deep continuous-valued neural networks require a fixed number of operations to achieve their classification error rate. From the examples of LeNet for MNIST and BinaryNet for CIFAR-10, we show that with an increase in error rate of a few percentage points, the SNNs can achieve more than 2x reductions in operations compared to the original CNNs. This highlights the potential of SNNs in particular when deployed on power-efficient neuromorphic spiking neuron chips, for use in embedded applications. PMID:29375284

Representational Distance Learning for Deep Neural Networks

PubMed Central

McClure, Patrick; Kriegeskorte, Nikolaus

2016-01-01

Deep neural networks (DNNs) provide useful models of visual representational transformations. We present a method that enables a DNN (student) to learn from the internal representational spaces of a reference model (teacher), which could be another DNN or, in the future, a biological brain. Representational spaces of the student and the teacher are characterized by representational distance matrices (RDMs). We propose representational distance learning (RDL), a stochastic gradient descent method that drives the RDMs of the student to approximate the RDMs of the teacher. We demonstrate that RDL is competitive with other transfer learning techniques for two publicly available benchmark computer vision datasets (MNIST and CIFAR-100), while allowing for architectural differences between student and teacher. By pulling the student's RDMs toward those of the teacher, RDL significantly improved visual classification performance when compared to baseline networks that did not use transfer learning. In the future, RDL may enable combined supervised training of deep neural networks using task constraints (e.g., images and category labels) and constraints from brain-activity measurements, so as to build models that replicate the internal representational spaces of biological brains. PMID:28082889

Representational Distance Learning for Deep Neural Networks.

PubMed

McClure, Patrick; Kriegeskorte, Nikolaus

2016-01-01

Deep neural networks (DNNs) provide useful models of visual representational transformations. We present a method that enables a DNN (student) to learn from the internal representational spaces of a reference model (teacher), which could be another DNN or, in the future, a biological brain. Representational spaces of the student and the teacher are characterized by representational distance matrices (RDMs). We propose representational distance learning (RDL), a stochastic gradient descent method that drives the RDMs of the student to approximate the RDMs of the teacher. We demonstrate that RDL is competitive with other transfer learning techniques for two publicly available benchmark computer vision datasets (MNIST and CIFAR-100), while allowing for architectural differences between student and teacher. By pulling the student's RDMs toward those of the teacher, RDL significantly improved visual classification performance when compared to baseline networks that did not use transfer learning. In the future, RDL may enable combined supervised training of deep neural networks using task constraints (e.g., images and category labels) and constraints from brain-activity measurements, so as to build models that replicate the internal representational spaces of biological brains.

Fast, Simple and Accurate Handwritten Digit Classification by Training Shallow Neural Network Classifiers with the ‘Extreme Learning Machine’ Algorithm

PubMed Central

McDonnell, Mark D.; Tissera, Migel D.; Vladusich, Tony; van Schaik, André; Tapson, Jonathan

2015-01-01

Recent advances in training deep (multi-layer) architectures have inspired a renaissance in neural network use. For example, deep convolutional networks are becoming the default option for difficult tasks on large datasets, such as image and speech recognition. However, here we show that error rates below 1% on the MNIST handwritten digit benchmark can be replicated with shallow non-convolutional neural networks. This is achieved by training such networks using the ‘Extreme Learning Machine’ (ELM) approach, which also enables a very rapid training time (∼ 10 minutes). Adding distortions, as is common practise for MNIST, reduces error rates even further. Our methods are also shown to be capable of achieving less than 5.5% error rates on the NORB image database. To achieve these results, we introduce several enhancements to the standard ELM algorithm, which individually and in combination can significantly improve performance. The main innovation is to ensure each hidden-unit operates only on a randomly sized and positioned patch of each image. This form of random ‘receptive field’ sampling of the input ensures the input weight matrix is sparse, with about 90% of weights equal to zero. Furthermore, combining our methods with a small number of iterations of a single-batch backpropagation method can significantly reduce the number of hidden-units required to achieve a particular performance. Our close to state-of-the-art results for MNIST and NORB suggest that the ease of use and accuracy of the ELM algorithm for designing a single-hidden-layer neural network classifier should cause it to be given greater consideration either as a standalone method for simpler problems, or as the final classification stage in deep neural networks applied to more difficult problems. PMID:26262687

DEEP MOTIF DASHBOARD: VISUALIZING AND UNDERSTANDING GENOMIC SEQUENCES USING DEEP NEURAL NETWORKS.

PubMed

Lanchantin, Jack; Singh, Ritambhara; Wang, Beilun; Qi, Yanjun

2017-01-01

Deep neural network (DNN) models have recently obtained state-of-the-art prediction accuracy for the transcription factor binding (TFBS) site classification task. However, it remains unclear how these approaches identify meaningful DNA sequence signals and give insights as to why TFs bind to certain locations. In this paper, we propose a toolkit called the Deep Motif Dashboard (DeMo Dashboard) which provides a suite of visualization strategies to extract motifs, or sequence patterns from deep neural network models for TFBS classification. We demonstrate how to visualize and understand three important DNN models: convolutional, recurrent, and convolutional-recurrent networks. Our first visualization method is finding a test sequence's saliency map which uses first-order derivatives to describe the importance of each nucleotide in making the final prediction. Second, considering recurrent models make predictions in a temporal manner (from one end of a TFBS sequence to the other), we introduce temporal output scores, indicating the prediction score of a model over time for a sequential input. Lastly, a class-specific visualization strategy finds the optimal input sequence for a given TFBS positive class via stochastic gradient optimization. Our experimental results indicate that a convolutional-recurrent architecture performs the best among the three architectures. The visualization techniques indicate that CNN-RNN makes predictions by modeling both motifs as well as dependencies among them.

Deep Motif Dashboard: Visualizing and Understanding Genomic Sequences Using Deep Neural Networks

PubMed Central

Lanchantin, Jack; Singh, Ritambhara; Wang, Beilun; Qi, Yanjun

2018-01-01

Deep neural network (DNN) models have recently obtained state-of-the-art prediction accuracy for the transcription factor binding (TFBS) site classification task. However, it remains unclear how these approaches identify meaningful DNA sequence signals and give insights as to why TFs bind to certain locations. In this paper, we propose a toolkit called the Deep Motif Dashboard (DeMo Dashboard) which provides a suite of visualization strategies to extract motifs, or sequence patterns from deep neural network models for TFBS classification. We demonstrate how to visualize and understand three important DNN models: convolutional, recurrent, and convolutional-recurrent networks. Our first visualization method is finding a test sequence’s saliency map which uses first-order derivatives to describe the importance of each nucleotide in making the final prediction. Second, considering recurrent models make predictions in a temporal manner (from one end of a TFBS sequence to the other), we introduce temporal output scores, indicating the prediction score of a model over time for a sequential input. Lastly, a class-specific visualization strategy finds the optimal input sequence for a given TFBS positive class via stochastic gradient optimization. Our experimental results indicate that a convolutional-recurrent architecture performs the best among the three architectures. The visualization techniques indicate that CNN-RNN makes predictions by modeling both motifs as well as dependencies among them. PMID:27896980

Deep Learning Automates the Quantitative Analysis of Individual Cells in Live-Cell Imaging Experiments.

PubMed

Van Valen, David A; Kudo, Takamasa; Lane, Keara M; Macklin, Derek N; Quach, Nicolas T; DeFelice, Mialy M; Maayan, Inbal; Tanouchi, Yu; Ashley, Euan A; Covert, Markus W

2016-11-01

Live-cell imaging has opened an exciting window into the role cellular heterogeneity plays in dynamic, living systems. A major critical challenge for this class of experiments is the problem of image segmentation, or determining which parts of a microscope image correspond to which individual cells. Current approaches require many hours of manual curation and depend on approaches that are difficult to share between labs. They are also unable to robustly segment the cytoplasms of mammalian cells. Here, we show that deep convolutional neural networks, a supervised machine learning method, can solve this challenge for multiple cell types across the domains of life. We demonstrate that this approach can robustly segment fluorescent images of cell nuclei as well as phase images of the cytoplasms of individual bacterial and mammalian cells from phase contrast images without the need for a fluorescent cytoplasmic marker. These networks also enable the simultaneous segmentation and identification of different mammalian cell types grown in co-culture. A quantitative comparison with prior methods demonstrates that convolutional neural networks have improved accuracy and lead to a significant reduction in curation time. We relay our experience in designing and optimizing deep convolutional neural networks for this task and outline several design rules that we found led to robust performance. We conclude that deep convolutional neural networks are an accurate method that require less curation time, are generalizable to a multiplicity of cell types, from bacteria to mammalian cells, and expand live-cell imaging capabilities to include multi-cell type systems.

Deep Learning Automates the Quantitative Analysis of Individual Cells in Live-Cell Imaging Experiments

DOE PAGES

Van Valen, David A.; Kudo, Takamasa; Lane, Keara M.; ...

2016-11-04

Live-cell imaging has opened an exciting window into the role cellular heterogeneity plays in dynamic, living systems. A major critical challenge for this class of experiments is the problem of image segmentation, or determining which parts of a microscope image correspond to which individual cells. Current approaches require many hours of manual curation and depend on approaches that are difficult to share between labs. They are also unable to robustly segment the cytoplasms of mammalian cells. Here, we show that deep convolutional neural networks, a supervised machine learning method, can solve this challenge for multiple cell types across the domainsmore » of life. We demonstrate that this approach can robustly segment fluorescent images of cell nuclei as well as phase images of the cytoplasms of individual bacterial and mammalian cells from phase contrast images without the need for a fluorescent cytoplasmic marker. These networks also enable the simultaneous segmentation and identification of different mammalian cell types grown in co-culture. A quantitative comparison with prior methods demonstrates that convolutional neural networks have improved accuracy and lead to a significant reduction in curation time. We relay our experience in designing and optimizing deep convolutional neural networks for this task and outline several design rules that we found led to robust performance. We conclude that deep convolutional neural networks are an accurate method that require less curation time, are generalizable to a multiplicity of cell types, from bacteria to mammalian cells, and expand live-cell imaging capabilities to include multi-cell type systems.« less

Deep Learning Automates the Quantitative Analysis of Individual Cells in Live-Cell Imaging Experiments

DOE Office of Scientific and Technical Information (OSTI.GOV)

Van Valen, David A.; Kudo, Takamasa; Lane, Keara M.

Live-cell imaging has opened an exciting window into the role cellular heterogeneity plays in dynamic, living systems. A major critical challenge for this class of experiments is the problem of image segmentation, or determining which parts of a microscope image correspond to which individual cells. Current approaches require many hours of manual curation and depend on approaches that are difficult to share between labs. They are also unable to robustly segment the cytoplasms of mammalian cells. Here, we show that deep convolutional neural networks, a supervised machine learning method, can solve this challenge for multiple cell types across the domainsmore » of life. We demonstrate that this approach can robustly segment fluorescent images of cell nuclei as well as phase images of the cytoplasms of individual bacterial and mammalian cells from phase contrast images without the need for a fluorescent cytoplasmic marker. These networks also enable the simultaneous segmentation and identification of different mammalian cell types grown in co-culture. A quantitative comparison with prior methods demonstrates that convolutional neural networks have improved accuracy and lead to a significant reduction in curation time. We relay our experience in designing and optimizing deep convolutional neural networks for this task and outline several design rules that we found led to robust performance. We conclude that deep convolutional neural networks are an accurate method that require less curation time, are generalizable to a multiplicity of cell types, from bacteria to mammalian cells, and expand live-cell imaging capabilities to include multi-cell type systems.« less

Deep Learning Automates the Quantitative Analysis of Individual Cells in Live-Cell Imaging Experiments

PubMed Central

Van Valen, David A.; Lane, Keara M.; Quach, Nicolas T.; Maayan, Inbal

2016-01-01

Live-cell imaging has opened an exciting window into the role cellular heterogeneity plays in dynamic, living systems. A major critical challenge for this class of experiments is the problem of image segmentation, or determining which parts of a microscope image correspond to which individual cells. Current approaches require many hours of manual curation and depend on approaches that are difficult to share between labs. They are also unable to robustly segment the cytoplasms of mammalian cells. Here, we show that deep convolutional neural networks, a supervised machine learning method, can solve this challenge for multiple cell types across the domains of life. We demonstrate that this approach can robustly segment fluorescent images of cell nuclei as well as phase images of the cytoplasms of individual bacterial and mammalian cells from phase contrast images without the need for a fluorescent cytoplasmic marker. These networks also enable the simultaneous segmentation and identification of different mammalian cell types grown in co-culture. A quantitative comparison with prior methods demonstrates that convolutional neural networks have improved accuracy and lead to a significant reduction in curation time. We relay our experience in designing and optimizing deep convolutional neural networks for this task and outline several design rules that we found led to robust performance. We conclude that deep convolutional neural networks are an accurate method that require less curation time, are generalizable to a multiplicity of cell types, from bacteria to mammalian cells, and expand live-cell imaging capabilities to include multi-cell type systems. PMID:27814364

Accurate Natural Trail Detection Using a Combination of a Deep Neural Network and Dynamic Programming.

PubMed

Adhikari, Shyam Prasad; Yang, Changju; Slot, Krzysztof; Kim, Hyongsuk

2018-01-10

This paper presents a vision sensor-based solution to the challenging problem of detecting and following trails in highly unstructured natural environments like forests, rural areas and mountains, using a combination of a deep neural network and dynamic programming. The deep neural network (DNN) concept has recently emerged as a very effective tool for processing vision sensor signals. A patch-based DNN is trained with supervised data to classify fixed-size image patches into "trail" and "non-trail" categories, and reshaped to a fully convolutional architecture to produce trail segmentation map for arbitrary-sized input images. As trail and non-trail patches do not exhibit clearly defined shapes or forms, the patch-based classifier is prone to misclassification, and produces sub-optimal trail segmentation maps. Dynamic programming is introduced to find an optimal trail on the sub-optimal DNN output map. Experimental results showing accurate trail detection for real-world trail datasets captured with a head mounted vision system are presented.

Deep Gaze Velocity Analysis During Mammographic Reading for Biometric Identification of Radiologists

DOE Office of Scientific and Technical Information (OSTI.GOV)

Yoon, Hong-Jun; Alamudun, Folami T.; Hudson, Kathy

Several studies have confirmed that the gaze velocity of the human eye can be utilized as a behavioral biometric or personalized biomarker. In this study, we leverage the local feature representation capacity of convolutional neural networks (CNNs) for eye gaze velocity analysis as the basis for biometric identification of radiologists performing breast cancer screening. Using gaze data collected from 10 radiologists reading 100 mammograms of various diagnoses, we compared the performance of a CNN-based classification algorithm with two deep learning classifiers, deep neural network and deep belief network, and a previously presented hidden Markov model classifier. The study showed thatmore » the CNN classifier is superior compared to alternative classification methods based on macro F1-scores derived from 10-fold cross-validation experiments. Our results further support the efficacy of eye gaze velocity as a biometric identifier of medical imaging experts.« less

Deep Gaze Velocity Analysis During Mammographic Reading for Biometric Identification of Radiologists

DOE PAGES

Yoon, Hong-Jun; Alamudun, Folami T.; Hudson, Kathy; ...

2018-01-24

Several studies have confirmed that the gaze velocity of the human eye can be utilized as a behavioral biometric or personalized biomarker. In this study, we leverage the local feature representation capacity of convolutional neural networks (CNNs) for eye gaze velocity analysis as the basis for biometric identification of radiologists performing breast cancer screening. Using gaze data collected from 10 radiologists reading 100 mammograms of various diagnoses, we compared the performance of a CNN-based classification algorithm with two deep learning classifiers, deep neural network and deep belief network, and a previously presented hidden Markov model classifier. The study showed thatmore » the CNN classifier is superior compared to alternative classification methods based on macro F1-scores derived from 10-fold cross-validation experiments. Our results further support the efficacy of eye gaze velocity as a biometric identifier of medical imaging experts.« less

The Livermore Brain: Massive Deep Learning Networks Enabled by High Performance Computing

DOE Office of Scientific and Technical Information (OSTI.GOV)

Chen, Barry Y.

The proliferation of inexpensive sensor technologies like the ubiquitous digital image sensors has resulted in the collection and sharing of vast amounts of unsorted and unexploited raw data. Companies and governments who are able to collect and make sense of large datasets to help them make better decisions more rapidly will have a competitive advantage in the information era. Machine Learning technologies play a critical role for automating the data understanding process; however, to be maximally effective, useful intermediate representations of the data are required. These representations or “features” are transformations of the raw data into a form where patternsmore » are more easily recognized. Recent breakthroughs in Deep Learning have made it possible to learn these features from large amounts of labeled data. The focus of this project is to develop and extend Deep Learning algorithms for learning features from vast amounts of unlabeled data and to develop the HPC neural network training platform to support the training of massive network models. This LDRD project succeeded in developing new unsupervised feature learning algorithms for images and video and created a scalable neural network training toolkit for HPC. Additionally, this LDRD helped create the world’s largest freely-available image and video dataset supporting open multimedia research and used this dataset for training our deep neural networks. This research helped LLNL capture several work-for-others (WFO) projects, attract new talent, and establish collaborations with leading academic and commercial partners. Finally, this project demonstrated the successful training of the largest unsupervised image neural network using HPC resources and helped establish LLNL leadership at the intersection of Machine Learning and HPC research.« less

Training Deep Convolutional Neural Networks with Resistive Cross-Point Devices

PubMed Central

Gokmen, Tayfun; Onen, Murat; Haensch, Wilfried

2017-01-01

In a previous work we have detailed the requirements for obtaining maximal deep learning performance benefit by implementing fully connected deep neural networks (DNN) in the form of arrays of resistive devices. Here we extend the concept of Resistive Processing Unit (RPU) devices to convolutional neural networks (CNNs). We show how to map the convolutional layers to fully connected RPU arrays such that the parallelism of the hardware can be fully utilized in all three cycles of the backpropagation algorithm. We find that the noise and bound limitations imposed by the analog nature of the computations performed on the arrays significantly affect the training accuracy of the CNNs. Noise and bound management techniques are presented that mitigate these problems without introducing any additional complexity in the analog circuits and that can be addressed by the digital circuits. In addition, we discuss digitally programmable update management and device variability reduction techniques that can be used selectively for some of the layers in a CNN. We show that a combination of all those techniques enables a successful application of the RPU concept for training CNNs. The techniques discussed here are more general and can be applied beyond CNN architectures and therefore enables applicability of the RPU approach to a large class of neural network architectures. PMID:29066942

Training Deep Convolutional Neural Networks with Resistive Cross-Point Devices.

PubMed

Gokmen, Tayfun; Onen, Murat; Haensch, Wilfried

2017-01-01

In a previous work we have detailed the requirements for obtaining maximal deep learning performance benefit by implementing fully connected deep neural networks (DNN) in the form of arrays of resistive devices. Here we extend the concept of Resistive Processing Unit (RPU) devices to convolutional neural networks (CNNs). We show how to map the convolutional layers to fully connected RPU arrays such that the parallelism of the hardware can be fully utilized in all three cycles of the backpropagation algorithm. We find that the noise and bound limitations imposed by the analog nature of the computations performed on the arrays significantly affect the training accuracy of the CNNs. Noise and bound management techniques are presented that mitigate these problems without introducing any additional complexity in the analog circuits and that can be addressed by the digital circuits. In addition, we discuss digitally programmable update management and device variability reduction techniques that can be used selectively for some of the layers in a CNN. We show that a combination of all those techniques enables a successful application of the RPU concept for training CNNs. The techniques discussed here are more general and can be applied beyond CNN architectures and therefore enables applicability of the RPU approach to a large class of neural network architectures.

Single image super-resolution based on convolutional neural networks

NASA Astrophysics Data System (ADS)

Zou, Lamei; Luo, Ming; Yang, Weidong; Li, Peng; Jin, Liujia

2018-03-01

We present a deep learning method for single image super-resolution (SISR). The proposed approach learns end-to-end mapping between low-resolution (LR) images and high-resolution (HR) images. The mapping is represented as a deep convolutional neural network which inputs the LR image and outputs the HR image. Our network uses 5 convolution layers, which kernels size include 5×5, 3×3 and 1×1. In our proposed network, we use residual-learning and combine different sizes of convolution kernels at the same layer. The experiment results show that our proposed method performs better than the existing methods in reconstructing quality index and human visual effects on benchmarked images.

Epithelium-Stroma Classification via Convolutional Neural Networks and Unsupervised Domain Adaptation in Histopathological Images.

PubMed

Huang, Yue; Zheng, Han; Liu, Chi; Ding, Xinghao; Rohde, Gustavo K

2017-11-01

Epithelium-stroma classification is a necessary preprocessing step in histopathological image analysis. Current deep learning based recognition methods for histology data require collection of large volumes of labeled data in order to train a new neural network when there are changes to the image acquisition procedure. However, it is extremely expensive for pathologists to manually label sufficient volumes of data for each pathology study in a professional manner, which results in limitations in real-world applications. A very simple but effective deep learning method, that introduces the concept of unsupervised domain adaptation to a simple convolutional neural network (CNN), has been proposed in this paper. Inspired by transfer learning, our paper assumes that the training data and testing data follow different distributions, and there is an adaptation operation to more accurately estimate the kernels in CNN in feature extraction, in order to enhance performance by transferring knowledge from labeled data in source domain to unlabeled data in target domain. The model has been evaluated using three independent public epithelium-stroma datasets by cross-dataset validations. The experimental results demonstrate that for epithelium-stroma classification, the proposed framework outperforms the state-of-the-art deep neural network model, and it also achieves better performance than other existing deep domain adaptation methods. The proposed model can be considered to be a better option for real-world applications in histopathological image analysis, since there is no longer a requirement for large-scale labeled data in each specified domain.

Modeling language and cognition with deep unsupervised learning: a tutorial overview

PubMed Central

Zorzi, Marco; Testolin, Alberto; Stoianov, Ivilin P.

2013-01-01

Deep unsupervised learning in stochastic recurrent neural networks with many layers of hidden units is a recent breakthrough in neural computation research. These networks build a hierarchy of progressively more complex distributed representations of the sensory data by fitting a hierarchical generative model. In this article we discuss the theoretical foundations of this approach and we review key issues related to training, testing and analysis of deep networks for modeling language and cognitive processing. The classic letter and word perception problem of McClelland and Rumelhart (1981) is used as a tutorial example to illustrate how structured and abstract representations may emerge from deep generative learning. We argue that the focus on deep architectures and generative (rather than discriminative) learning represents a crucial step forward for the connectionist modeling enterprise, because it offers a more plausible model of cortical learning as well as a way to bridge the gap between emergentist connectionist models and structured Bayesian models of cognition. PMID:23970869

Modeling language and cognition with deep unsupervised learning: a tutorial overview.

PubMed

Zorzi, Marco; Testolin, Alberto; Stoianov, Ivilin P

2013-01-01

Deep unsupervised learning in stochastic recurrent neural networks with many layers of hidden units is a recent breakthrough in neural computation research. These networks build a hierarchy of progressively more complex distributed representations of the sensory data by fitting a hierarchical generative model. In this article we discuss the theoretical foundations of this approach and we review key issues related to training, testing and analysis of deep networks for modeling language and cognitive processing. The classic letter and word perception problem of McClelland and Rumelhart (1981) is used as a tutorial example to illustrate how structured and abstract representations may emerge from deep generative learning. We argue that the focus on deep architectures and generative (rather than discriminative) learning represents a crucial step forward for the connectionist modeling enterprise, because it offers a more plausible model of cortical learning as well as a way to bridge the gap between emergentist connectionist models and structured Bayesian models of cognition.

Action-Driven Visual Object Tracking With Deep Reinforcement Learning.

PubMed

Yun, Sangdoo; Choi, Jongwon; Yoo, Youngjoon; Yun, Kimin; Choi, Jin Young

2018-06-01

In this paper, we propose an efficient visual tracker, which directly captures a bounding box containing the target object in a video by means of sequential actions learned using deep neural networks. The proposed deep neural network to control tracking actions is pretrained using various training video sequences and fine-tuned during actual tracking for online adaptation to a change of target and background. The pretraining is done by utilizing deep reinforcement learning (RL) as well as supervised learning. The use of RL enables even partially labeled data to be successfully utilized for semisupervised learning. Through the evaluation of the object tracking benchmark data set, the proposed tracker is validated to achieve a competitive performance at three times the speed of existing deep network-based trackers. The fast version of the proposed method, which operates in real time on graphics processing unit, outperforms the state-of-the-art real-time trackers with an accuracy improvement of more than 8%.

Revealing Fundamental Physics from the Daya Bay Neutrino Experiment Using Deep Neural Networks

DOE PAGES

Racah, Evan; Ko, Seyoon; Sadowski, Peter; ...

2017-02-02

Experiments in particle physics produce enormous quantities of data that must be analyzed and interpreted by teams of physicists. This analysis is often exploratory, where scientists are unable to enumerate the possible types of signal prior to performing the experiment. Thus, tools for summarizing, clustering, visualizing and classifying high-dimensional data are essential. Here in this work, we show that meaningful physical content can be revealed by transforming the raw data into a learned high-level representation using deep neural networks, with measurements taken at the Daya Bay Neutrino Experiment as a case study. We further show how convolutional deep neural networksmore » can provide an effective classification filter with greater than 97% accuracy across different classes of physics events, significantly better than other machine learning approaches.« less

Hidden Markov models and neural networks for fault detection in dynamic systems

NASA Technical Reports Server (NTRS)

Smyth, Padhraic

1994-01-01

Neural networks plus hidden Markov models (HMM) can provide excellent detection and false alarm rate performance in fault detection applications, as shown in this viewgraph presentation. Modified models allow for novelty detection. Key contributions of neural network models are: (1) excellent nonparametric discrimination capability; (2) a good estimator of posterior state probabilities, even in high dimensions, and thus can be embedded within overall probabilistic model (HMM); and (3) simple to implement compared to other nonparametric models. Neural network/HMM monitoring model is currently being integrated with the new Deep Space Network (DSN) antenna controller software and will be on-line monitoring a new DSN 34-m antenna (DSS-24) by July, 1994.

Deep Convolutional and LSTM Recurrent Neural Networks for Multimodal Wearable Activity Recognition

PubMed Central

Ordóñez, Francisco Javier; Roggen, Daniel

2016-01-01

Human activity recognition (HAR) tasks have traditionally been solved using engineered features obtained by heuristic processes. Current research suggests that deep convolutional neural networks are suited to automate feature extraction from raw sensor inputs. However, human activities are made of complex sequences of motor movements, and capturing this temporal dynamics is fundamental for successful HAR. Based on the recent success of recurrent neural networks for time series domains, we propose a generic deep framework for activity recognition based on convolutional and LSTM recurrent units, which: (i) is suitable for multimodal wearable sensors; (ii) can perform sensor fusion naturally; (iii) does not require expert knowledge in designing features; and (iv) explicitly models the temporal dynamics of feature activations. We evaluate our framework on two datasets, one of which has been used in a public activity recognition challenge. Our results show that our framework outperforms competing deep non-recurrent networks on the challenge dataset by 4% on average; outperforming some of the previous reported results by up to 9%. Our results show that the framework can be applied to homogeneous sensor modalities, but can also fuse multimodal sensors to improve performance. We characterise key architectural hyperparameters’ influence on performance to provide insights about their optimisation. PMID:26797612

Deep neural networks for texture classification-A theoretical analysis.

PubMed

Basu, Saikat; Mukhopadhyay, Supratik; Karki, Manohar; DiBiano, Robert; Ganguly, Sangram; Nemani, Ramakrishna; Gayaka, Shreekant

2018-01-01

We investigate the use of Deep Neural Networks for the classification of image datasets where texture features are important for generating class-conditional discriminative representations. To this end, we first derive the size of the feature space for some standard textural features extracted from the input dataset and then use the theory of Vapnik-Chervonenkis dimension to show that hand-crafted feature extraction creates low-dimensional representations which help in reducing the overall excess error rate. As a corollary to this analysis, we derive for the first time upper bounds on the VC dimension of Convolutional Neural Network as well as Dropout and Dropconnect networks and the relation between excess error rate of Dropout and Dropconnect networks. The concept of intrinsic dimension is used to validate the intuition that texture-based datasets are inherently higher dimensional as compared to handwritten digits or other object recognition datasets and hence more difficult to be shattered by neural networks. We then derive the mean distance from the centroid to the nearest and farthest sampling points in an n-dimensional manifold and show that the Relative Contrast of the sample data vanishes as dimensionality of the underlying vector space tends to infinity. Copyright © 2017 Elsevier Ltd. All rights reserved.

Deep learning with convolutional neural network in radiology.

PubMed

Yasaka, Koichiro; Akai, Hiroyuki; Kunimatsu, Akira; Kiryu, Shigeru; Abe, Osamu

2018-04-01

Deep learning with a convolutional neural network (CNN) is gaining attention recently for its high performance in image recognition. Images themselves can be utilized in a learning process with this technique, and feature extraction in advance of the learning process is not required. Important features can be automatically learned. Thanks to the development of hardware and software in addition to techniques regarding deep learning, application of this technique to radiological images for predicting clinically useful information, such as the detection and the evaluation of lesions, etc., are beginning to be investigated. This article illustrates basic technical knowledge regarding deep learning with CNNs along the actual course (collecting data, implementing CNNs, and training and testing phases). Pitfalls regarding this technique and how to manage them are also illustrated. We also described some advanced topics of deep learning, results of recent clinical studies, and the future directions of clinical application of deep learning techniques.

Collaborative identification method for sea battlefield target based on deep convolutional neural networks

NASA Astrophysics Data System (ADS)

Zheng, Guangdi; Pan, Mingbo; Liu, Wei; Wu, Xuetong

2018-03-01

The target identification of the sea battlefield is the prerequisite for the judgment of the enemy in the modern naval battle. In this paper, a collaborative identification method based on convolution neural network is proposed to identify the typical targets of sea battlefields. Different from the traditional single-input/single-output identification method, the proposed method constructs a multi-input/single-output co-identification architecture based on optimized convolution neural network and weighted D-S evidence theory. The simulation results show that

Deep learning in color: towards automated quark/gluon jet discrimination

DOE PAGES

Komiske, Patrick T.; Metodiev, Eric M.; Schwartz, Matthew D.

2017-01-25

Artificial intelligence offers the potential to automate challenging data-processing tasks in collider physics. Here, to establish its prospects, we explore to what extent deep learning with convolutional neural networks can discriminate quark and gluon jets better than observables designed by physicists. Our approach builds upon the paradigm that a jet can be treated as an image, with intensity given by the local calorimeter deposits. We supplement this construction by adding color to the images, with red, green and blue intensities given by the transverse momentum in charged particles, transverse momentum in neutral particles, and pixel-level charged particle counts. Overall, themore » deep networks match or outperform traditional jet variables. We also find that, while various simulations produce different quark and gluon jets, the neural networks are surprisingly insensitive to these differences, similar to traditional observables. This suggests that the networks can extract robust physical information from imperfect simulations.« less

Deep learning in color: towards automated quark/gluon jet discrimination

NASA Astrophysics Data System (ADS)

Komiske, Patrick T.; Metodiev, Eric M.; Schwartz, Matthew D.

2017-01-01

Artificial intelligence offers the potential to automate challenging data-processing tasks in collider physics. To establish its prospects, we explore to what extent deep learning with convolutional neural networks can discriminate quark and gluon jets better than observables designed by physicists. Our approach builds upon the paradigm that a jet can be treated as an image, with intensity given by the local calorimeter deposits. We supplement this construction by adding color to the images, with red, green and blue intensities given by the transverse momentum in charged particles, transverse momentum in neutral particles, and pixel-level charged particle counts. Overall, the deep networks match or outperform traditional jet variables. We also find that, while various simulations produce different quark and gluon jets, the neural networks are surprisingly insensitive to these differences, similar to traditional observables. This suggests that the networks can extract robust physical information from imperfect simulations.

Deep learning in color: towards automated quark/gluon jet discrimination

DOE Office of Scientific and Technical Information (OSTI.GOV)

Komiske, Patrick T.; Metodiev, Eric M.; Schwartz, Matthew D.

Artificial intelligence offers the potential to automate challenging data-processing tasks in collider physics. Here, to establish its prospects, we explore to what extent deep learning with convolutional neural networks can discriminate quark and gluon jets better than observables designed by physicists. Our approach builds upon the paradigm that a jet can be treated as an image, with intensity given by the local calorimeter deposits. We supplement this construction by adding color to the images, with red, green and blue intensities given by the transverse momentum in charged particles, transverse momentum in neutral particles, and pixel-level charged particle counts. Overall, themore » deep networks match or outperform traditional jet variables. We also find that, while various simulations produce different quark and gluon jets, the neural networks are surprisingly insensitive to these differences, similar to traditional observables. This suggests that the networks can extract robust physical information from imperfect simulations.« less

Automated Deep Learning-Based System to Identify Endothelial Cells Derived from Induced Pluripotent Stem Cells.

PubMed

Kusumoto, Dai; Lachmann, Mark; Kunihiro, Takeshi; Yuasa, Shinsuke; Kishino, Yoshikazu; Kimura, Mai; Katsuki, Toshiomi; Itoh, Shogo; Seki, Tomohisa; Fukuda, Keiichi

2018-06-05

Deep learning technology is rapidly advancing and is now used to solve complex problems. Here, we used deep learning in convolutional neural networks to establish an automated method to identify endothelial cells derived from induced pluripotent stem cells (iPSCs), without the need for immunostaining or lineage tracing. Networks were trained to predict whether phase-contrast images contain endothelial cells based on morphology only. Predictions were validated by comparison to immunofluorescence staining for CD31, a marker of endothelial cells. Method parameters were then automatically and iteratively optimized to increase prediction accuracy. We found that prediction accuracy was correlated with network depth and pixel size of images to be analyzed. Finally, K-fold cross-validation confirmed that optimized convolutional neural networks can identify endothelial cells with high performance, based only on morphology. Copyright © 2018 The Author(s). Published by Elsevier Inc. All rights reserved.

Deep Learning and Its Applications in Biomedicine.

PubMed

Cao, Chensi; Liu, Feng; Tan, Hai; Song, Deshou; Shu, Wenjie; Li, Weizhong; Zhou, Yiming; Bo, Xiaochen; Xie, Zhi

2018-02-01

Advances in biological and medical technologies have been providing us explosive volumes of biological and physiological data, such as medical images, electroencephalography, genomic and protein sequences. Learning from these data facilitates the understanding of human health and disease. Developed from artificial neural networks, deep learning-based algorithms show great promise in extracting features and learning patterns from complex data. The aim of this paper is to provide an overview of deep learning techniques and some of the state-of-the-art applications in the biomedical field. We first introduce the development of artificial neural network and deep learning. We then describe two main components of deep learning, i.e., deep learning architectures and model optimization. Subsequently, some examples are demonstrated for deep learning applications, including medical image classification, genomic sequence analysis, as well as protein structure classification and prediction. Finally, we offer our perspectives for the future directions in the field of deep learning. Copyright © 2018. Production and hosting by Elsevier B.V.

Solving differential equations with unknown constitutive relations as recurrent neural networks

DOE Office of Scientific and Technical Information (OSTI.GOV)

Hagge, Tobias J.; Stinis, Panagiotis; Yeung, Enoch H.

We solve a system of ordinary differential equations with an unknown functional form of a sink (reaction rate) term. We assume that the measurements (time series) of state variables are partially available, and use a recurrent neural network to “learn” the reaction rate from this data. This is achieved by including discretized ordinary differential equations as part of a recurrent neural network training problem. We extend TensorFlow’s recurrent neural network architecture to create a simple but scalable and effective solver for the unknown functions, and apply it to a fedbatch bioreactor simulation problem. Use of techniques from recent deep learningmore » literature enables training of functions with behavior manifesting over thousands of time steps. Our networks are structurally similar to recurrent neural networks, but differ in purpose, and require modified training strategies.« less

Automatic recognition of holistic functional brain networks using iteratively optimized convolutional neural networks (IO-CNN) with weak label initialization.

PubMed

Zhao, Yu; Ge, Fangfei; Liu, Tianming

2018-07-01

fMRI data decomposition techniques have advanced significantly from shallow models such as Independent Component Analysis (ICA) and Sparse Coding and Dictionary Learning (SCDL) to deep learning models such Deep Belief Networks (DBN) and Convolutional Autoencoder (DCAE). However, interpretations of those decomposed networks are still open questions due to the lack of functional brain atlases, no correspondence across decomposed or reconstructed networks across different subjects, and significant individual variabilities. Recent studies showed that deep learning, especially deep convolutional neural networks (CNN), has extraordinary ability of accommodating spatial object patterns, e.g., our recent works using 3D CNN for fMRI-derived network classifications achieved high accuracy with a remarkable tolerance for mistakenly labelled training brain networks. However, the training data preparation is one of the biggest obstacles in these supervised deep learning models for functional brain network map recognitions, since manual labelling requires tedious and time-consuming labours which will sometimes even introduce label mistakes. Especially for mapping functional networks in large scale datasets such as hundreds of thousands of brain networks used in this paper, the manual labelling method will become almost infeasible. In response, in this work, we tackled both the network recognition and training data labelling tasks by proposing a new iteratively optimized deep learning CNN (IO-CNN) framework with an automatic weak label initialization, which enables the functional brain networks recognition task to a fully automatic large-scale classification procedure. Our extensive experiments based on ABIDE-II 1099 brains' fMRI data showed the great promise of our IO-CNN framework. Copyright © 2018 Elsevier B.V. All rights reserved.

Squeeze-SegNet: a new fast deep convolutional neural network for semantic segmentation

NASA Astrophysics Data System (ADS)

Nanfack, Geraldin; Elhassouny, Azeddine; Oulad Haj Thami, Rachid

2018-04-01

The recent researches in Deep Convolutional Neural Network have focused their attention on improving accuracy that provide significant advances. However, if they were limited to classification tasks, nowadays with contributions from Scientific Communities who are embarking in this field, they have become very useful in higher level tasks such as object detection and pixel-wise semantic segmentation. Thus, brilliant ideas in the field of semantic segmentation with deep learning have completed the state of the art of accuracy, however this architectures become very difficult to apply in embedded systems as is the case for autonomous driving. We present a new Deep fully Convolutional Neural Network for pixel-wise semantic segmentation which we call Squeeze-SegNet. The architecture is based on Encoder-Decoder style. We use a SqueezeNet-like encoder and a decoder formed by our proposed squeeze-decoder module and upsample layer using downsample indices like in SegNet and we add a deconvolution layer to provide final multi-channel feature map. On datasets like Camvid or City-states, our net gets SegNet-level accuracy with less than 10 times fewer parameters than SegNet.

Evaluation of a deep learning architecture for MR imaging prediction of ATRX in glioma patients

NASA Astrophysics Data System (ADS)

Korfiatis, Panagiotis; Kline, Timothy L.; Erickson, Bradley J.

2018-02-01

Predicting mutation/loss of alpha-thalassemia/mental retardation syndrome X-linked (ATRX) gene utilizing MR imaging is of high importance since it is a predictor of response and prognosis in brain tumors. In this study, we compare a deep neural network approach based on a residual deep neural network (ResNet) architecture and one based on a classical machine learning approach and evaluate their ability in predicting ATRX mutation status without the need for a distinct tumor segmentation step. We found that the ResNet50 (50 layers) architecture, pre trained on ImageNet data was the best performing model, achieving an accuracy of 0.91 for the test set (classification of a slice as no tumor, ATRX mutated, or mutated) in terms of f1 score in a test set of 35 cases. The SVM classifier achieved 0.63 for differentiating the Flair signal abnormality regions from the test patients based on their mutation status. We report a method that alleviates the need for extensive preprocessing and acts as a proof of concept that deep neural network architectures can be used to predict molecular biomarkers from routine medical images.

Visualizing deep neural network by alternately image blurring and deblurring.

PubMed

Wang, Feng; Liu, Haijun; Cheng, Jian

2018-01-01

Visualization from trained deep neural networks has drawn massive public attention in recent. One of the visualization approaches is to train images maximizing the activation of specific neurons. However, directly maximizing the activation would lead to unrecognizable images, which cannot provide any meaningful information. In this paper, we introduce a simple but effective technique to constrain the optimization route of the visualization. By adding two totally inverse transformations, image blurring and deblurring, to the optimization procedure, recognizable images can be created. Our algorithm is good at extracting the details in the images, which are usually filtered by previous methods in the visualizations. Extensive experiments on AlexNet, VGGNet and GoogLeNet illustrate that we can better understand the neural networks utilizing the knowledge obtained by the visualization. Copyright © 2017 Elsevier Ltd. All rights reserved.

Deep Gate Recurrent Neural Network

DTIC Science & Technology

2016-11-22

Schmidhuber. A system for robotic heart surgery that learns to tie knots using recurrent neural networks. In IEEE International Conference on...tasks, such as Machine Translation (Bahdanau et al. (2015)) or Robot Reinforcement Learning (Bakker (2001)). The main idea behind these networks is to...and J. Peters. Reinforcement learning in robotics : A survey. The International Journal of Robotics Research, 32:1238–1274, 2013. ISSN 0278-3649. doi

Speaker-dependent Multipitch Tracking Using Deep Neural Networks

DTIC Science & Technology

2015-01-01

connections through time. Studies have shown that RNNs are good at modeling sequential data like handwriting [12] and speech [26]. We plan to explore RNNs in...Schmidhuber, and S. Fernández, “Unconstrained on-line handwriting recognition with recurrent neural networks,” in Proceedings of NIPS, 2008, pp. 577–584. [13

Construction of multi-agent mobile robots control system in the problem of persecution with using a modified reinforcement learning method based on neural networks

NASA Astrophysics Data System (ADS)

Patkin, M. L.; Rogachev, G. N.

2018-02-01

A method for constructing a multi-agent control system for mobile robots based on training with reinforcement using deep neural networks is considered. Synthesis of the management system is proposed to be carried out with reinforcement training and the modified Actor-Critic method, in which the Actor module is divided into Action Actor and Communication Actor in order to simultaneously manage mobile robots and communicate with partners. Communication is carried out by sending partners at each step a vector of real numbers that are added to the observation vector and affect the behaviour. Functions of Actors and Critic are approximated by deep neural networks. The Critics value function is trained by using the TD-error method and the Actor’s function by using DDPG. The Communication Actor’s neural network is trained through gradients received from partner agents. An environment in which a cooperative multi-agent interaction is present was developed, computer simulation of the application of this method in the control problem of two robots pursuing two goals was carried out.

Deep learning and the electronic structure problem

NASA Astrophysics Data System (ADS)

Mills, Kyle; Spanner, Michael; Tamblyn, Isaac

In the past decade, the fields of artificial intelligence and computer vision have progressed remarkably. Supported by the enthusiasm of large tech companies, as well as significant hardware advances and the utilization of graphical processing units to accelerate computations, deep neural networks (DNN) are gaining momentum as a robust choice for many diverse machine learning applications. We have demonstrated the ability of a DNN to solve a quantum mechanical eigenvalue equation directly, without the need to compute a wavefunction, and without knowledge of the underlying physics. We have trained a convolutional neural network to predict the total energy of an electron in a confining, 2-dimensional electrostatic potential. We numerically solved the one-electron Schrödinger equation for millions of electrostatic potentials, and used this as training data for our neural network. Four classes of potentials were assessed: the canonical cases of the harmonic oscillator and infinite well, and two types of randomly generated potentials for which no analytic solution is known. We compare the performance of the neural network and consider how these results could lead to future advances in electronic structure theory.

Cardiac Arrhythmia Classification by Multi-Layer Perceptron and Convolution Neural Networks.

PubMed

Savalia, Shalin; Emamian, Vahid

2018-05-04

The electrocardiogram (ECG) plays an imperative role in the medical field, as it records heart signal over time and is used to discover numerous cardiovascular diseases. If a documented ECG signal has a certain irregularity in its predefined features, this is called arrhythmia, the types of which include tachycardia, bradycardia, supraventricular arrhythmias, and ventricular, etc. This has encouraged us to do research that consists of distinguishing between several arrhythmias by using deep neural network algorithms such as multi-layer perceptron (MLP) and convolution neural network (CNN). The TensorFlow library that was established by Google for deep learning and machine learning is used in python to acquire the algorithms proposed here. The ECG databases accessible at PhysioBank.com and kaggle.com were used for training, testing, and validation of the MLP and CNN algorithms. The proposed algorithm consists of four hidden layers with weights, biases in MLP, and four-layer convolution neural networks which map ECG samples to the different classes of arrhythmia. The accuracy of the algorithm surpasses the performance of the current algorithms that have been developed by other cardiologists in both sensitivity and precision.

Geometrical structure of Neural Networks: Geodesics, Jeffrey's Prior and Hyper-ribbons

NASA Astrophysics Data System (ADS)

Hayden, Lorien; Alemi, Alex; Sethna, James

2014-03-01

Neural networks are learning algorithms which are employed in a host of Machine Learning problems including speech recognition, object classification and data mining. In practice, neural networks learn a low dimensional representation of high dimensional data and define a model manifold which is an embedding of this low dimensional structure in the higher dimensional space. In this work, we explore the geometrical structure of a neural network model manifold. A Stacked Denoising Autoencoder and a Deep Belief Network are trained on handwritten digits from the MNIST database. Construction of geodesics along the surface and of slices taken from the high dimensional manifolds reveal a hierarchy of widths corresponding to a hyper-ribbon structure. This property indicates that neural networks fall into the class of sloppy models, in which certain parameter combinations dominate the behavior. Employing this information could prove valuable in designing both neural network architectures and training algorithms. This material is based upon work supported by the National Science Foundation Graduate Research Fellowship under Grant No . DGE-1144153.

A Multiobjective Sparse Feature Learning Model for Deep Neural Networks.

PubMed

Gong, Maoguo; Liu, Jia; Li, Hao; Cai, Qing; Su, Linzhi

2015-12-01

Hierarchical deep neural networks are currently popular learning models for imitating the hierarchical architecture of human brain. Single-layer feature extractors are the bricks to build deep networks. Sparse feature learning models are popular models that can learn useful representations. But most of those models need a user-defined constant to control the sparsity of representations. In this paper, we propose a multiobjective sparse feature learning model based on the autoencoder. The parameters of the model are learnt by optimizing two objectives, reconstruction error and the sparsity of hidden units simultaneously to find a reasonable compromise between them automatically. We design a multiobjective induced learning procedure for this model based on a multiobjective evolutionary algorithm. In the experiments, we demonstrate that the learning procedure is effective, and the proposed multiobjective model can learn useful sparse features.

Deep neural networks to enable real-time multimessenger astrophysics

NASA Astrophysics Data System (ADS)

George, Daniel; Huerta, E. A.

2018-02-01

Gravitational wave astronomy has set in motion a scientific revolution. To further enhance the science reach of this emergent field of research, there is a pressing need to increase the depth and speed of the algorithms used to enable these ground-breaking discoveries. We introduce Deep Filtering—a new scalable machine learning method for end-to-end time-series signal processing. Deep Filtering is based on deep learning with two deep convolutional neural networks, which are designed for classification and regression, to detect gravitational wave signals in highly noisy time-series data streams and also estimate the parameters of their sources in real time. Acknowledging that some of the most sensitive algorithms for the detection of gravitational waves are based on implementations of matched filtering, and that a matched filter is the optimal linear filter in Gaussian noise, the application of Deep Filtering using whitened signals in Gaussian noise is investigated in this foundational article. The results indicate that Deep Filtering outperforms conventional machine learning techniques, achieves similar performance compared to matched filtering, while being several orders of magnitude faster, allowing real-time signal processing with minimal resources. Furthermore, we demonstrate that Deep Filtering can detect and characterize waveform signals emitted from new classes of eccentric or spin-precessing binary black holes, even when trained with data sets of only quasicircular binary black hole waveforms. The results presented in this article, and the recent use of deep neural networks for the identification of optical transients in telescope data, suggests that deep learning can facilitate real-time searches of gravitational wave sources and their electromagnetic and astroparticle counterparts. In the subsequent article, the framework introduced herein is directly applied to identify and characterize gravitational wave events in real LIGO data.

Large deep neural networks for MS lesion segmentation

NASA Astrophysics Data System (ADS)

Prieto, Juan C.; Cavallari, Michele; Palotai, Miklos; Morales Pinzon, Alfredo; Egorova, Svetlana; Styner, Martin; Guttmann, Charles R. G.

2017-02-01

Multiple sclerosis (MS) is a multi-factorial autoimmune disorder, characterized by spatial and temporal dissemination of brain lesions that are visible in T2-weighted and Proton Density (PD) MRI. Assessment of lesion burden and is useful for monitoring the course of the disease, and assessing correlates of clinical outcomes. Although there are established semi-automated methods to measure lesion volume, most of them require human interaction and editing, which are time consuming and limits the ability to analyze large sets of data with high accuracy. The primary objective of this work is to improve existing segmentation algorithms and accelerate the time consuming operation of identifying and validating MS lesions. In this paper, a Deep Neural Network for MS Lesion Segmentation is implemented. The MS lesion samples are extracted from the Partners Comprehensive Longitudinal Investigation of Multiple Sclerosis (CLIMB) study. A set of 900 subjects with T2, PD and a manually corrected label map images were used to train a Deep Neural Network and identify MS lesions. Initial tests using this network achieved a 90% accuracy rate. A secondary goal was to enable this data repository for big data analysis by using this algorithm to segment the remaining cases available in the CLIMB repository.

Improving protein disorder prediction by deep bidirectional long short-term memory recurrent neural networks.

PubMed

Hanson, Jack; Yang, Yuedong; Paliwal, Kuldip; Zhou, Yaoqi

2017-03-01

Capturing long-range interactions between structural but not sequence neighbors of proteins is a long-standing challenging problem in bioinformatics. Recently, long short-term memory (LSTM) networks have significantly improved the accuracy of speech and image classification problems by remembering useful past information in long sequential events. Here, we have implemented deep bidirectional LSTM recurrent neural networks in the problem of protein intrinsic disorder prediction. The new method, named SPOT-Disorder, has steadily improved over a similar method using a traditional, window-based neural network (SPINE-D) in all datasets tested without separate training on short and long disordered regions. Independent tests on four other datasets including the datasets from critical assessment of structure prediction (CASP) techniques and >10 000 annotated proteins from MobiDB, confirmed SPOT-Disorder as one of the best methods in disorder prediction. Moreover, initial studies indicate that the method is more accurate in predicting functional sites in disordered regions. These results highlight the usefulness combining LSTM with deep bidirectional recurrent neural networks in capturing non-local, long-range interactions for bioinformatics applications. SPOT-disorder is available as a web server and as a standalone program at: http://sparks-lab.org/server/SPOT-disorder/index.php . j.hanson@griffith.edu.au or yuedong.yang@griffith.edu.au or yaoqi.zhou@griffith.edu.au. Supplementary data is available at Bioinformatics online. © The Author 2016. Published by Oxford University Press. All rights reserved. For Permissions, please e-mail: journals.permissions@oup.com

Imbalance aware lithography hotspot detection: a deep learning approach

NASA Astrophysics Data System (ADS)

Yang, Haoyu; Luo, Luyang; Su, Jing; Lin, Chenxi; Yu, Bei

2017-03-01

With the advancement of VLSI technology nodes, light diffraction caused lithographic hotspots have become a serious problem affecting manufacture yield. Lithography hotspot detection at the post-OPC stage is imperative to check potential circuit failures when transferring designed patterns onto silicon wafers. Although conventional lithography hotspot detection methods, such as machine learning, have gained satisfactory performance, with extreme scaling of transistor feature size and more and more complicated layout patterns, conventional methodologies may suffer from performance degradation. For example, manual or ad hoc feature extraction in a machine learning framework may lose important information when predicting potential errors in ultra-large-scale integrated circuit masks. In this paper, we present a deep convolutional neural network (CNN) targeting representative feature learning in lithography hotspot detection. We carefully analyze impact and effectiveness of different CNN hyper-parameters, through which a hotspot-detection-oriented neural network model is established. Because hotspot patterns are always minorities in VLSI mask design, the training data set is highly imbalanced. In this situation, a neural network is no longer reliable, because a trained model with high classification accuracy may still suffer from high false negative results (missing hotspots), which is fatal in hotspot detection problems. To address the imbalance problem, we further apply minority upsampling and random-mirror flipping before training the network. Experimental results show that our proposed neural network model achieves highly comparable or better performance on the ICCAD 2012 contest benchmark compared to state-of-the-art hotspot detectors based on deep or representative machine leaning.

Reducing weight precision of convolutional neural networks towards large-scale on-chip image recognition

NASA Astrophysics Data System (ADS)

Ji, Zhengping; Ovsiannikov, Ilia; Wang, Yibing; Shi, Lilong; Zhang, Qiang

2015-05-01

In this paper, we develop a server-client quantization scheme to reduce bit resolution of deep learning architecture, i.e., Convolutional Neural Networks, for image recognition tasks. Low bit resolution is an important factor in bringing the deep learning neural network into hardware implementation, which directly determines the cost and power consumption. We aim to reduce the bit resolution of the network without sacrificing its performance. To this end, we design a new quantization algorithm called supervised iterative quantization to reduce the bit resolution of learned network weights. In the training stage, the supervised iterative quantization is conducted via two steps on server - apply k-means based adaptive quantization on learned network weights and retrain the network based on quantized weights. These two steps are alternated until the convergence criterion is met. In this testing stage, the network configuration and low-bit weights are loaded to the client hardware device to recognize coming input in real time, where optimized but expensive quantization becomes infeasible. Considering this, we adopt a uniform quantization for the inputs and internal network responses (called feature maps) to maintain low on-chip expenses. The Convolutional Neural Network with reduced weight and input/response precision is demonstrated in recognizing two types of images: one is hand-written digit images and the other is real-life images in office scenarios. Both results show that the new network is able to achieve the performance of the neural network with full bit resolution, even though in the new network the bit resolution of both weight and input are significantly reduced, e.g., from 64 bits to 4-5 bits.

Brain tumor segmentation in multi-spectral MRI using convolutional neural networks (CNN).

PubMed

Iqbal, Sajid; Ghani, M Usman; Saba, Tanzila; Rehman, Amjad

2018-04-01

A tumor could be found in any area of the brain and could be of any size, shape, and contrast. There may exist multiple tumors of different types in a human brain at the same time. Accurate tumor area segmentation is considered primary step for treatment of brain tumors. Deep Learning is a set of promising techniques that could provide better results as compared to nondeep learning techniques for segmenting timorous part inside a brain. This article presents a deep convolutional neural network (CNN) to segment brain tumors in MRIs. The proposed network uses BRATS segmentation challenge dataset which is composed of images obtained through four different modalities. Accordingly, we present an extended version of existing network to solve segmentation problem. The network architecture consists of multiple neural network layers connected in sequential order with the feeding of Convolutional feature maps at the peer level. Experimental results on BRATS 2015 benchmark data thus show the usability of the proposed approach and its superiority over the other approaches in this area of research. © 2018 Wiley Periodicals, Inc.

End-to-End Multimodal Emotion Recognition Using Deep Neural Networks

NASA Astrophysics Data System (ADS)

Tzirakis, Panagiotis; Trigeorgis, George; Nicolaou, Mihalis A.; Schuller, Bjorn W.; Zafeiriou, Stefanos

2017-12-01

Automatic affect recognition is a challenging task due to the various modalities emotions can be expressed with. Applications can be found in many domains including multimedia retrieval and human computer interaction. In recent years, deep neural networks have been used with great success in determining emotional states. Inspired by this success, we propose an emotion recognition system using auditory and visual modalities. To capture the emotional content for various styles of speaking, robust features need to be extracted. To this purpose, we utilize a Convolutional Neural Network (CNN) to extract features from the speech, while for the visual modality a deep residual network (ResNet) of 50 layers. In addition to the importance of feature extraction, a machine learning algorithm needs also to be insensitive to outliers while being able to model the context. To tackle this problem, Long Short-Term Memory (LSTM) networks are utilized. The system is then trained in an end-to-end fashion where - by also taking advantage of the correlations of the each of the streams - we manage to significantly outperform the traditional approaches based on auditory and visual handcrafted features for the prediction of spontaneous and natural emotions on the RECOLA database of the AVEC 2016 research challenge on emotion recognition.

Coarse-to-fine deep neural network for fast pedestrian detection

NASA Astrophysics Data System (ADS)

Li, Yaobin; Yang, Xinmei; Cao, Lijun

2017-11-01

Pedestrian detection belongs to a category of object detection is a key issue in the field of video surveillance and automatic driving. Although recent object detection methods, such as Fast/Faster RCNN, have achieved excellent performance, it is difficult to meet real-time requirements and limits the application in real scenarios. A coarse-to-fine deep neural network for fast pedestrian detection is proposed in this paper. Two-stage approach is presented to realize fine trade-off between accuracy and speed. In the coarse stage, we train a fast deep convolution neural network to generate most pedestrian candidates at the cost of a number of false positives. The detector can cover the majority of scales, sizes, and occlusions of pedestrians. After that, a classification network is introduced to refine the pedestrian candidates generated from the previous stage. Refining through classification network, most of false detections will be excluded easily and the final pedestrian predictions with bounding box and confidence score are produced. Competitive results have been achieved on INRIA dataset in terms of accuracy, especially the method can achieve real-time detection that is faster than the previous leading methods. The effectiveness of coarse-to-fine approach to detect pedestrians is verified, and the accuracy and stability are also improved.

A patch-based convolutional neural network for remote sensing image classification.

PubMed

Sharma, Atharva; Liu, Xiuwen; Yang, Xiaojun; Shi, Di

2017-11-01

Availability of accurate land cover information over large areas is essential to the global environment sustainability; digital classification using medium-resolution remote sensing data would provide an effective method to generate the required land cover information. However, low accuracy of existing per-pixel based classification methods for medium-resolution data is a fundamental limiting factor. While convolutional neural networks (CNNs) with deep layers have achieved unprecedented improvements in object recognition applications that rely on fine image structures, they cannot be applied directly to medium-resolution data due to lack of such fine structures. In this paper, considering the spatial relation of a pixel to its neighborhood, we propose a new deep patch-based CNN system tailored for medium-resolution remote sensing data. The system is designed by incorporating distinctive characteristics of medium-resolution data; in particular, the system computes patch-based samples from multidimensional top of atmosphere reflectance data. With a test site from the Florida Everglades area (with a size of 771 square kilometers), the proposed new system has outperformed pixel-based neural network, pixel-based CNN and patch-based neural network by 24.36%, 24.23% and 11.52%, respectively, in overall classification accuracy. By combining the proposed deep CNN and the huge collection of medium-resolution remote sensing data, we believe that much more accurate land cover datasets can be produced over large areas. Copyright © 2017 Elsevier Ltd. All rights reserved.

A new bio-inspired stimulator to suppress hyper-synchronized neural firing in a cortical network.

PubMed

Amiri, Masoud; Amiri, Mahmood; Nazari, Soheila; Faez, Karim

2016-12-07

Hyper-synchronous neural oscillations are the character of several neurological diseases such as epilepsy. On the other hand, glial cells and particularly astrocytes can influence neural synchronization. Therefore, based on the recent researches, a new bio-inspired stimulator is proposed which basically is a dynamical model of the astrocyte biophysical model. The performance of the new stimulator is investigated on a large-scale, cortical network. Both excitatory and inhibitory synapses are also considered in the simulated spiking neural network. The simulation results show that the new stimulator has a good performance and is able to reduce recurrent abnormal excitability which in turn avoids the hyper-synchronous neural firing in the spiking neural network. In this way, the proposed stimulator has a demand controlled characteristic and is a good candidate for deep brain stimulation (DBS) technique to successfully suppress the neural hyper-synchronization. Copyright © 2016 Elsevier Ltd. All rights reserved.

Recurrent neural networks for breast lesion classification based on DCE-MRIs

NASA Astrophysics Data System (ADS)

Antropova, Natasha; Huynh, Benjamin; Giger, Maryellen

2018-02-01

Dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) plays a significant role in breast cancer screening, cancer staging, and monitoring response to therapy. Recently, deep learning methods are being rapidly incorporated in image-based breast cancer diagnosis and prognosis. However, most of the current deep learning methods make clinical decisions based on 2-dimentional (2D) or 3D images and are not well suited for temporal image data. In this study, we develop a deep learning methodology that enables integration of clinically valuable temporal components of DCE-MRIs into deep learning-based lesion classification. Our work is performed on a database of 703 DCE-MRI cases for the task of distinguishing benign and malignant lesions, and uses the area under the ROC curve (AUC) as the performance metric in conducting that task. We train a recurrent neural network, specifically a long short-term memory network (LSTM), on sequences of image features extracted from the dynamic MRI sequences. These features are extracted with VGGNet, a convolutional neural network pre-trained on a large dataset of natural images ImageNet. The features are obtained from various levels of the network, to capture low-, mid-, and high-level information about the lesion. Compared to a classification method that takes as input only images at a single time-point (yielding an AUC = 0.81 (se = 0.04)), our LSTM method improves lesion classification with an AUC of 0.85 (se = 0.03).

White blood cells identification system based on convolutional deep neural learning networks.

PubMed

Shahin, A I; Guo, Yanhui; Amin, K M; Sharawi, Amr A

2017-11-16

White blood cells (WBCs) differential counting yields valued information about human health and disease. The current developed automated cell morphology equipments perform differential count which is based on blood smear image analysis. Previous identification systems for WBCs consist of successive dependent stages; pre-processing, segmentation, feature extraction, feature selection, and classification. There is a real need to employ deep learning methodologies so that the performance of previous WBCs identification systems can be increased. Classifying small limited datasets through deep learning systems is a major challenge and should be investigated. In this paper, we propose a novel identification system for WBCs based on deep convolutional neural networks. Two methodologies based on transfer learning are followed: transfer learning based on deep activation features and fine-tuning of existed deep networks. Deep acrivation featues are extracted from several pre-trained networks and employed in a traditional identification system. Moreover, a novel end-to-end convolutional deep architecture called "WBCsNet" is proposed and built from scratch. Finally, a limited balanced WBCs dataset classification is performed through the WBCsNet as a pre-trained network. During our experiments, three different public WBCs datasets (2551 images) have been used which contain 5 healthy WBCs types. The overall system accuracy achieved by the proposed WBCsNet is (96.1%) which is more than different transfer learning approaches or even the previous traditional identification system. We also present features visualization for the WBCsNet activation which reflects higher response than the pre-trained activated one. a novel WBCs identification system based on deep learning theory is proposed and a high performance WBCsNet can be employed as a pre-trained network. Copyright © 2017. Published by Elsevier B.V.

Comment on 'Deep convolutional neural network with transfer learning for rectum toxicity prediction in cervical cancer radiotherapy: a feasibility study'.

PubMed

Valdes, Gilmer; Interian, Yannet

2018-03-15

The application of machine learning (ML) presents tremendous opportunities for the field of oncology, thus we read 'Deep convolutional neural network with transfer learning for rectum toxicity prediction in cervical cancer radiotherapy: a feasibility study' with great interest. In this article, the authors used state of the art techniques: a pre-trained convolutional neural network (VGG-16 CNN), transfer learning, data augmentation, drop out and early stopping, all of which are directly responsible for the success and the excitement that these algorithms have created in other fields. We believe that the use of these techniques can offer tremendous opportunities in the field of Medical Physics and as such we would like to praise the authors for their pioneering application to the field of Radiation Oncology. That being said, given that the field of Medical Physics has unique characteristics that differentiate us from those fields where these techniques have been applied successfully, we would like to raise some points for future discussion and follow up studies that could help the community understand the limitations and nuances of deep learning techniques.

A convolutional neural network-based screening tool for X-ray serial crystallography

PubMed Central

Ke, Tsung-Wei; Brewster, Aaron S.; Yu, Stella X.; Ushizima, Daniela; Yang, Chao; Sauter, Nicholas K.

2018-01-01

A new tool is introduced for screening macromolecular X-ray crystallography diffraction images produced at an X-ray free-electron laser light source. Based on a data-driven deep learning approach, the proposed tool executes a convolutional neural network to detect Bragg spots. Automatic image processing algorithms described can enable the classification of large data sets, acquired under realistic conditions consisting of noisy data with experimental artifacts. Outcomes are compared for different data regimes, including samples from multiple instruments and differing amounts of training data for neural network optimization. PMID:29714177

A convolutional neural network-based screening tool for X-ray serial crystallography.

PubMed

Ke, Tsung Wei; Brewster, Aaron S; Yu, Stella X; Ushizima, Daniela; Yang, Chao; Sauter, Nicholas K

2018-05-01

A new tool is introduced for screening macromolecular X-ray crystallography diffraction images produced at an X-ray free-electron laser light source. Based on a data-driven deep learning approach, the proposed tool executes a convolutional neural network to detect Bragg spots. Automatic image processing algorithms described can enable the classification of large data sets, acquired under realistic conditions consisting of noisy data with experimental artifacts. Outcomes are compared for different data regimes, including samples from multiple instruments and differing amounts of training data for neural network optimization. open access.

A convolutional neural network-based screening tool for X-ray serial crystallography

DOE PAGES

Ke, Tsung-Wei; Brewster, Aaron S.; Yu, Stella X.; ...

2018-04-24

A new tool is introduced for screening macromolecular X-ray crystallography diffraction images produced at an X-ray free-electron laser light source. Based on a data-driven deep learning approach, the proposed tool executes a convolutional neural network to detect Bragg spots. Automatic image processing algorithms described can enable the classification of large data sets, acquired under realistic conditions consisting of noisy data with experimental artifacts. Outcomes are compared for different data regimes, including samples from multiple instruments and differing amounts of training data for neural network optimization.

A convolutional neural network-based screening tool for X-ray serial crystallography

DOE Office of Scientific and Technical Information (OSTI.GOV)

Ke, Tsung-Wei; Brewster, Aaron S.; Yu, Stella X.

A new tool is introduced for screening macromolecular X-ray crystallography diffraction images produced at an X-ray free-electron laser light source. Based on a data-driven deep learning approach, the proposed tool executes a convolutional neural network to detect Bragg spots. Automatic image processing algorithms described can enable the classification of large data sets, acquired under realistic conditions consisting of noisy data with experimental artifacts. Outcomes are compared for different data regimes, including samples from multiple instruments and differing amounts of training data for neural network optimization.

MUFOLD-SS: New deep inception-inside-inception networks for protein secondary structure prediction.

PubMed

Fang, Chao; Shang, Yi; Xu, Dong

2018-05-01

Protein secondary structure prediction can provide important information for protein 3D structure prediction and protein functions. Deep learning offers a new opportunity to significantly improve prediction accuracy. In this article, a new deep neural network architecture, named the Deep inception-inside-inception (Deep3I) network, is proposed for protein secondary structure prediction and implemented as a software tool MUFOLD-SS. The input to MUFOLD-SS is a carefully designed feature matrix corresponding to the primary amino acid sequence of a protein, which consists of a rich set of information derived from individual amino acid, as well as the context of the protein sequence. Specifically, the feature matrix is a composition of physio-chemical properties of amino acids, PSI-BLAST profile, and HHBlits profile. MUFOLD-SS is composed of a sequence of nested inception modules and maps the input matrix to either eight states or three states of secondary structures. The architecture of MUFOLD-SS enables effective processing of local and global interactions between amino acids in making accurate prediction. In extensive experiments on multiple datasets, MUFOLD-SS outperformed the best existing methods and other deep neural networks significantly. MUFold-SS can be downloaded from http://dslsrv8.cs.missouri.edu/~cf797/MUFoldSS/download.html. © 2018 Wiley Periodicals, Inc.

Deep-cascade: Cascading 3D Deep Neural Networks for Fast Anomaly Detection and Localization in Crowded Scenes.

PubMed

Sabokrou, Mohammad; Fayyaz, Mohsen; Fathy, Mahmood; Klette, Reinhard

2017-02-17

This paper proposes a fast and reliable method for anomaly detection and localization in video data showing crowded scenes. Time-efficient anomaly localization is an ongoing challenge and subject of this paper. We propose a cubicpatch- based method, characterised by a cascade of classifiers, which makes use of an advanced feature-learning approach. Our cascade of classifiers has two main stages. First, a light but deep 3D auto-encoder is used for early identification of "many" normal cubic patches. This deep network operates on small cubic patches as being the first stage, before carefully resizing remaining candidates of interest, and evaluating those at the second stage using a more complex and deeper 3D convolutional neural network (CNN). We divide the deep autoencoder and the CNN into multiple sub-stages which operate as cascaded classifiers. Shallow layers of the cascaded deep networks (designed as Gaussian classifiers, acting as weak single-class classifiers) detect "simple" normal patches such as background patches, and more complex normal patches are detected at deeper layers. It is shown that the proposed novel technique (a cascade of two cascaded classifiers) performs comparable to current top-performing detection and localization methods on standard benchmarks, but outperforms those in general with respect to required computation time.

Deep neural network-based bandwidth enhancement of photoacoustic data.

PubMed

Gutta, Sreedevi; Kadimesetty, Venkata Suryanarayana; Kalva, Sandeep Kumar; Pramanik, Manojit; Ganapathy, Sriram; Yalavarthy, Phaneendra K

2017-11-01

Photoacoustic (PA) signals collected at the boundary of tissue are always band-limited. A deep neural network was proposed to enhance the bandwidth (BW) of the detected PA signal, thereby improving the quantitative accuracy of the reconstructed PA images. A least square-based deconvolution method that utilizes the Tikhonov regularization framework was used for comparison with the proposed network. The proposed method was evaluated using both numerical and experimental data. The results indicate that the proposed method was capable of enhancing the BW of the detected PA signal, which inturn improves the contrast recovery and quality of reconstructed PA images without adding any significant computational burden. (2017) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE).

End-to-end deep neural network for optical inversion in quantitative photoacoustic imaging.

PubMed

Cai, Chuangjian; Deng, Kexin; Ma, Cheng; Luo, Jianwen

2018-06-15

An end-to-end deep neural network, ResU-net, is developed for quantitative photoacoustic imaging. A residual learning framework is used to facilitate optimization and to gain better accuracy from considerably increased network depth. The contracting and expanding paths enable ResU-net to extract comprehensive context information from multispectral initial pressure images and, subsequently, to infer a quantitative image of chromophore concentration or oxygen saturation (sO 2 ). According to our numerical experiments, the estimations of sO 2 and indocyanine green concentration are accurate and robust against variations in both optical property and object geometry. An extremely short reconstruction time of 22 ms is achieved.

An optimization method for speech enhancement based on deep neural network

NASA Astrophysics Data System (ADS)

Sun, Haixia; Li, Sikun

2017-06-01

Now, this document puts forward a deep neural network (DNN) model with more credible data set and more robust structure. First, we take two regularization skills, dropout and sparsity constraint to strengthen the generalization ability of the model. In this way, not only the model is able to reach the consistency between the pre-training model and the fine-tuning model, but also it reduce resource consumption. Then network compression by weights sharing and quantization is allowed to reduce storage cost. In the end, we evaluate the quality of the reconstructed speech according to different criterion. The result proofs that the improved framework has good performance on speech enhancement and meets the requirement of speech processing.

Application of structured support vector machine backpropagation to a convolutional neural network for human pose estimation.

PubMed

Witoonchart, Peerajak; Chongstitvatana, Prabhas

2017-08-01

In this study, for the first time, we show how to formulate a structured support vector machine (SSVM) as two layers in a convolutional neural network, where the top layer is a loss augmented inference layer and the bottom layer is the normal convolutional layer. We show that a deformable part model can be learned with the proposed structured SVM neural network by backpropagating the error of the deformable part model to the convolutional neural network. The forward propagation calculates the loss augmented inference and the backpropagation calculates the gradient from the loss augmented inference layer to the convolutional layer. Thus, we obtain a new type of convolutional neural network called an Structured SVM convolutional neural network, which we applied to the human pose estimation problem. This new neural network can be used as the final layers in deep learning. Our method jointly learns the structural model parameters and the appearance model parameters. We implemented our method as a new layer in the existing Caffe library. Copyright © 2017 Elsevier Ltd. All rights reserved.

Cytopathological image analysis using deep-learning networks in microfluidic microscopy.

PubMed

Gopakumar, G; Hari Babu, K; Mishra, Deepak; Gorthi, Sai Siva; Sai Subrahmanyam, Gorthi R K

2017-01-01

Cytopathologic testing is one of the most critical steps in the diagnosis of diseases, including cancer. However, the task is laborious and demands skill. Associated high cost and low throughput drew considerable interest in automating the testing process. Several neural network architectures were designed to provide human expertise to machines. In this paper, we explore and propose the feasibility of using deep-learning networks for cytopathologic analysis by performing the classification of three important unlabeled, unstained leukemia cell lines (K562, MOLT, and HL60). The cell images used in the classification are captured using a low-cost, high-throughput cell imaging technique: microfluidics-based imaging flow cytometry. We demonstrate that without any conventional fine segmentation followed by explicit feature extraction, the proposed deep-learning algorithms effectively classify the coarsely localized cell lines. We show that the designed deep belief network as well as the deeply pretrained convolutional neural network outperform the conventionally used decision systems and are important in the medical domain, where the availability of labeled data is limited for training. We hope that our work enables the development of a clinically significant high-throughput microfluidic microscopy-based tool for disease screening/triaging, especially in resource-limited settings.

Automatic detection and segmentation of brain metastases on multimodal MR images with a deep convolutional neural network.

PubMed

Charron, Odelin; Lallement, Alex; Jarnet, Delphine; Noblet, Vincent; Clavier, Jean-Baptiste; Meyer, Philippe

2018-04-01

Stereotactic treatments are today the reference techniques for the irradiation of brain metastases in radiotherapy. The dose per fraction is very high, and delivered in small volumes (diameter <1 cm). As part of these treatments, effective detection and precise segmentation of lesions are imperative. Many methods based on deep-learning approaches have been developed for the automatic segmentation of gliomas, but very little for that of brain metastases. We adapted an existing 3D convolutional neural network (DeepMedic) to detect and segment brain metastases on MRI. At first, we sought to adapt the network parameters to brain metastases. We then explored the single or combined use of different MRI modalities, by evaluating network performance in terms of detection and segmentation. We also studied the interest of increasing the database with virtual patients or of using an additional database in which the active parts of the metastases are separated from the necrotic parts. Our results indicated that a deep network approach is promising for the detection and the segmentation of brain metastases on multimodal MRI. Copyright © 2018 Elsevier Ltd. All rights reserved.

Detection of prostate cancer on multiparametric MRI

NASA Astrophysics Data System (ADS)

Seah, Jarrel C. Y.; Tang, Jennifer S. N.; Kitchen, Andy

2017-03-01

In this manuscript, we describe our approach and methods to the ProstateX challenge, which achieved an overall AUC of 0.84 and the runner-up position. We train a deep convolutional neural network to classify lesions marked on multiparametric MRI of the prostate as clinically significant or not. We implement a novel addition to the standard convolutional architecture described as auto-windowing which is clinically inspired and designed to overcome some of the difficulties faced in MRI interpretation, where high dynamic ranges and low contrast edges may cause difficulty for traditional convolutional neural networks trained on high contrast natural imagery. We demonstrate that this system can be trained end to end and outperforms a similar architecture without such additions. Although a relatively small training set was provided, we use extensive data augmentation to prevent overfitting and transfer learning to improve convergence speed, showing that deep convolutional neural networks can be feasibly trained on small datasets.

An Ensemble Deep Convolutional Neural Network Model with Improved D-S Evidence Fusion for Bearing Fault Diagnosis.

PubMed

Li, Shaobo; Liu, Guokai; Tang, Xianghong; Lu, Jianguang; Hu, Jianjun

2017-07-28

Intelligent machine health monitoring and fault diagnosis are becoming increasingly important for modern manufacturing industries. Current fault diagnosis approaches mostly depend on expert-designed features for building prediction models. In this paper, we proposed IDSCNN, a novel bearing fault diagnosis algorithm based on ensemble deep convolutional neural networks and an improved Dempster-Shafer theory based evidence fusion. The convolutional neural networks take the root mean square (RMS) maps from the FFT (Fast Fourier Transformation) features of the vibration signals from two sensors as inputs. The improved D-S evidence theory is implemented via distance matrix from evidences and modified Gini Index. Extensive evaluations of the IDSCNN on the Case Western Reserve Dataset showed that our IDSCNN algorithm can achieve better fault diagnosis performance than existing machine learning methods by fusing complementary or conflicting evidences from different models and sensors and adapting to different load conditions.

Deep convolutional neural network for the automated detection and diagnosis of seizure using EEG signals.

PubMed

Acharya, U Rajendra; Oh, Shu Lih; Hagiwara, Yuki; Tan, Jen Hong; Adeli, Hojjat

2017-09-27

An encephalogram (EEG) is a commonly used ancillary test to aide in the diagnosis of epilepsy. The EEG signal contains information about the electrical activity of the brain. Traditionally, neurologists employ direct visual inspection to identify epileptiform abnormalities. This technique can be time-consuming, limited by technical artifact, provides variable results secondary to reader expertise level, and is limited in identifying abnormalities. Therefore, it is essential to develop a computer-aided diagnosis (CAD) system to automatically distinguish the class of these EEG signals using machine learning techniques. This is the first study to employ the convolutional neural network (CNN) for analysis of EEG signals. In this work, a 13-layer deep convolutional neural network (CNN) algorithm is implemented to detect normal, preictal, and seizure classes. The proposed technique achieved an accuracy, specificity, and sensitivity of 88.67%, 90.00% and 95.00%, respectively. Copyright © 2017 Elsevier Ltd. All rights reserved.

An Ensemble Deep Convolutional Neural Network Model with Improved D-S Evidence Fusion for Bearing Fault Diagnosis

PubMed Central

Li, Shaobo; Liu, Guokai; Tang, Xianghong; Lu, Jianguang

2017-01-01

Intelligent machine health monitoring and fault diagnosis are becoming increasingly important for modern manufacturing industries. Current fault diagnosis approaches mostly depend on expert-designed features for building prediction models. In this paper, we proposed IDSCNN, a novel bearing fault diagnosis algorithm based on ensemble deep convolutional neural networks and an improved Dempster–Shafer theory based evidence fusion. The convolutional neural networks take the root mean square (RMS) maps from the FFT (Fast Fourier Transformation) features of the vibration signals from two sensors as inputs. The improved D-S evidence theory is implemented via distance matrix from evidences and modified Gini Index. Extensive evaluations of the IDSCNN on the Case Western Reserve Dataset showed that our IDSCNN algorithm can achieve better fault diagnosis performance than existing machine learning methods by fusing complementary or conflicting evidences from different models and sensors and adapting to different load conditions. PMID:28788099

Deep Convolutional Neural Networks for Classifying Body Constitution Based on Face Image.

PubMed

Huan, Er-Yang; Wen, Gui-Hua; Zhang, Shi-Jun; Li, Dan-Yang; Hu, Yang; Chang, Tian-Yuan; Wang, Qing; Huang, Bing-Lin

2017-01-01

Body constitution classification is the basis and core content of traditional Chinese medicine constitution research. It is to extract the relevant laws from the complex constitution phenomenon and finally build the constitution classification system. Traditional identification methods have the disadvantages of inefficiency and low accuracy, for instance, questionnaires. This paper proposed a body constitution recognition algorithm based on deep convolutional neural network, which can classify individual constitution types according to face images. The proposed model first uses the convolutional neural network to extract the features of face image and then combines the extracted features with the color features. Finally, the fusion features are input to the Softmax classifier to get the classification result. Different comparison experiments show that the algorithm proposed in this paper can achieve the accuracy of 65.29% about the constitution classification. And its performance was accepted by Chinese medicine practitioners.

Deep learning-based features of breast MRI for prediction of occult invasive disease following a diagnosis of ductal carcinoma in situ: preliminary data

NASA Astrophysics Data System (ADS)

Zhu, Zhe; Harowicz, Michael; Zhang, Jun; Saha, Ashirbani; Grimm, Lars J.; Hwang, Shelley; Mazurowski, Maciej A.

2018-02-01

Approximately 25% of patients with ductal carcinoma in situ (DCIS) diagnosed from core needle biopsy are subsequently upstaged to invasive cancer at surgical excision. Identifying patients with occult invasive disease is important as it changes treatment and precludes enrollment in active surveillance for DCIS. In this study, we investigated upstaging of DCIS to invasive disease using deep features. While deep neural networks require large amounts of training data, the available data to predict DCIS upstaging is sparse and thus directly training a neural network is unlikely to be successful. In this work, a pre-trained neural network is used as a feature extractor and a support vector machine (SVM) is trained on the extracted features. We used the dynamic contrast-enhanced (DCE) MRIs of patients at our institution from January 1, 2000, through March 23, 2014 who underwent MRI following a diagnosis of DCIS. Among the 131 DCIS patients, there were 35 patients who were upstaged to invasive cancer. Area under the ROC curve within the 10-fold cross-validation scheme was used for validation of our predictive model. The use of deep features was able to achieve an AUC of 0.68 (95% CI: 0.56-0.78) to predict occult invasive disease. This preliminary work demonstrates the promise of deep features to predict surgical upstaging following a diagnosis of DCIS.

Self-learning Monte Carlo with deep neural networks

NASA Astrophysics Data System (ADS)

Shen, Huitao; Liu, Junwei; Fu, Liang

2018-05-01

The self-learning Monte Carlo (SLMC) method is a general algorithm to speedup MC simulations. Its efficiency has been demonstrated in various systems by introducing an effective model to propose global moves in the configuration space. In this paper, we show that deep neural networks can be naturally incorporated into SLMC, and without any prior knowledge can learn the original model accurately and efficiently. Demonstrated in quantum impurity models, we reduce the complexity for a local update from O (β2) in Hirsch-Fye algorithm to O (β lnβ ) , which is a significant speedup especially for systems at low temperatures.

Intelligent fault diagnosis of rolling bearings using an improved deep recurrent neural network

NASA Astrophysics Data System (ADS)

Jiang, Hongkai; Li, Xingqiu; Shao, Haidong; Zhao, Ke

2018-06-01

Traditional intelligent fault diagnosis methods for rolling bearings heavily depend on manual feature extraction and feature selection. For this purpose, an intelligent deep learning method, named the improved deep recurrent neural network (DRNN), is proposed in this paper. Firstly, frequency spectrum sequences are used as inputs to reduce the input size and ensure good robustness. Secondly, DRNN is constructed by the stacks of the recurrent hidden layer to automatically extract the features from the input spectrum sequences. Thirdly, an adaptive learning rate is adopted to improve the training performance of the constructed DRNN. The proposed method is verified with experimental rolling bearing data, and the results confirm that the proposed method is more effective than traditional intelligent fault diagnosis methods.

Incorporating deep learning with convolutional neural networks and position specific scoring matrices for identifying electron transport proteins.

PubMed

Le, Nguyen-Quoc-Khanh; Ho, Quang-Thai; Ou, Yu-Yen

2017-09-05

In several years, deep learning is a modern machine learning technique using in a variety of fields with state-of-the-art performance. Therefore, utilization of deep learning to enhance performance is also an important solution for current bioinformatics field. In this study, we try to use deep learning via convolutional neural networks and position specific scoring matrices to identify electron transport proteins, which is an important molecular function in transmembrane proteins. Our deep learning method can approach a precise model for identifying of electron transport proteins with achieved sensitivity of 80.3%, specificity of 94.4%, and accuracy of 92.3%, with MCC of 0.71 for independent dataset. The proposed technique can serve as a powerful tool for identifying electron transport proteins and can help biologists understand the function of the electron transport proteins. Moreover, this study provides a basis for further research that can enrich a field of applying deep learning in bioinformatics. © 2017 Wiley Periodicals, Inc. © 2017 Wiley Periodicals, Inc.

pDeep: Predicting MS/MS Spectra of Peptides with Deep Learning.

PubMed

Zhou, Xie-Xuan; Zeng, Wen-Feng; Chi, Hao; Luo, Chunjie; Liu, Chao; Zhan, Jianfeng; He, Si-Min; Zhang, Zhifei

2017-12-05

In tandem mass spectrometry (MS/MS)-based proteomics, search engines rely on comparison between an experimental MS/MS spectrum and the theoretical spectra of the candidate peptides. Hence, accurate prediction of the theoretical spectra of peptides appears to be particularly important. Here, we present pDeep, a deep neural network-based model for the spectrum prediction of peptides. Using the bidirectional long short-term memory (BiLSTM), pDeep can predict higher-energy collisional dissociation, electron-transfer dissociation, and electron-transfer and higher-energy collision dissociation MS/MS spectra of peptides with >0.9 median Pearson correlation coefficients. Further, we showed that intermediate layer of the neural network could reveal physicochemical properties of amino acids, for example the similarities of fragmentation behaviors between amino acids. We also showed the potential of pDeep to distinguish extremely similar peptides (peptides that contain isobaric amino acids, for example, GG = N, AG = Q, or even I = L), which were very difficult to distinguish using traditional search engines.

PIV-DCNN: cascaded deep convolutional neural networks for particle image velocimetry

NASA Astrophysics Data System (ADS)

Lee, Yong; Yang, Hua; Yin, Zhouping

2017-12-01

Velocity estimation (extracting the displacement vector information) from the particle image pairs is of critical importance for particle image velocimetry. This problem is mostly transformed into finding the sub-pixel peak in a correlation map. To address the original displacement extraction problem, we propose a different evaluation scheme (PIV-DCNN) with four-level regression deep convolutional neural networks. At each level, the networks are trained to predict a vector from two input image patches. The low-level network is skilled at large displacement estimation and the high- level networks are devoted to improving the accuracy. Outlier replacement and symmetric window offset operation glue the well- functioning networks in a cascaded manner. Through comparison with the standard PIV methods (one-pass cross-correlation method, three-pass window deformation), the practicability of the proposed PIV-DCNN is verified by the application to a diversity of synthetic and experimental PIV images.

Decoding of finger trajectory from ECoG using deep learning.

PubMed

Xie, Ziqian; Schwartz, Odelia; Prasad, Abhishek

2018-06-01

Conventional decoding pipeline for brain-machine interfaces (BMIs) consists of chained different stages of feature extraction, time-frequency analysis and statistical learning models. Each of these stages uses a different algorithm trained in a sequential manner, which makes it difficult to make the whole system adaptive. The goal was to create an adaptive online system with a single objective function and a single learning algorithm so that the whole system can be trained in parallel to increase the decoding performance. Here, we used deep neural networks consisting of convolutional neural networks (CNN) and a special kind of recurrent neural network (RNN) called long short term memory (LSTM) to address these needs. We used electrocorticography (ECoG) data collected by Kubanek et al. The task consisted of individual finger flexions upon a visual cue. Our model combined a hierarchical feature extractor CNN and a RNN that was able to process sequential data and recognize temporal dynamics in the neural data. CNN was used as the feature extractor and LSTM was used as the regression algorithm to capture the temporal dynamics of the signal. We predicted the finger trajectory using ECoG signals and compared results for the least angle regression (LARS), CNN-LSTM, random forest, LSTM model (LSTM_HC, for using hard-coded features) and a decoding pipeline consisting of band-pass filtering, energy extraction, feature selection and linear regression. The results showed that the deep learning models performed better than the commonly used linear model. The deep learning models not only gave smoother and more realistic trajectories but also learned the transition between movement and rest state. This study demonstrated a decoding network for BMI that involved a convolutional and recurrent neural network model. It integrated the feature extraction pipeline into the convolution and pooling layer and used LSTM layer to capture the state transitions. The discussed network eliminated the need to separately train the model at each step in the decoding pipeline. The whole system can be jointly optimized using stochastic gradient descent and is capable of online learning.

Decoding of finger trajectory from ECoG using deep learning

NASA Astrophysics Data System (ADS)

Xie, Ziqian; Schwartz, Odelia; Prasad, Abhishek

2018-06-01

Objective. Conventional decoding pipeline for brain-machine interfaces (BMIs) consists of chained different stages of feature extraction, time-frequency analysis and statistical learning models. Each of these stages uses a different algorithm trained in a sequential manner, which makes it difficult to make the whole system adaptive. The goal was to create an adaptive online system with a single objective function and a single learning algorithm so that the whole system can be trained in parallel to increase the decoding performance. Here, we used deep neural networks consisting of convolutional neural networks (CNN) and a special kind of recurrent neural network (RNN) called long short term memory (LSTM) to address these needs. Approach. We used electrocorticography (ECoG) data collected by Kubanek et al. The task consisted of individual finger flexions upon a visual cue. Our model combined a hierarchical feature extractor CNN and a RNN that was able to process sequential data and recognize temporal dynamics in the neural data. CNN was used as the feature extractor and LSTM was used as the regression algorithm to capture the temporal dynamics of the signal. Main results. We predicted the finger trajectory using ECoG signals and compared results for the least angle regression (LARS), CNN-LSTM, random forest, LSTM model (LSTM_HC, for using hard-coded features) and a decoding pipeline consisting of band-pass filtering, energy extraction, feature selection and linear regression. The results showed that the deep learning models performed better than the commonly used linear model. The deep learning models not only gave smoother and more realistic trajectories but also learned the transition between movement and rest state. Significance. This study demonstrated a decoding network for BMI that involved a convolutional and recurrent neural network model. It integrated the feature extraction pipeline into the convolution and pooling layer and used LSTM layer to capture the state transitions. The discussed network eliminated the need to separately train the model at each step in the decoding pipeline. The whole system can be jointly optimized using stochastic gradient descent and is capable of online learning.

Ship detection in optical remote sensing images based on deep convolutional neural networks

NASA Astrophysics Data System (ADS)

Yao, Yuan; Jiang, Zhiguo; Zhang, Haopeng; Zhao, Danpei; Cai, Bowen

2017-10-01

Automatic ship detection in optical remote sensing images has attracted wide attention for its broad applications. Major challenges for this task include the interference of cloud, wave, wake, and the high computational expenses. We propose a fast and robust ship detection algorithm to solve these issues. The framework for ship detection is designed based on deep convolutional neural networks (CNNs), which provide the accurate locations of ship targets in an efficient way. First, the deep CNN is designed to extract features. Then, a region proposal network (RPN) is applied to discriminate ship targets and regress the detection bounding boxes, in which the anchors are designed by intrinsic shape of ship targets. Experimental results on numerous panchromatic images demonstrate that, in comparison with other state-of-the-art ship detection methods, our method is more efficient and achieves higher detection accuracy and more precise bounding boxes in different complex backgrounds.

A light and faster regional convolutional neural network for object detection in optical remote sensing images

NASA Astrophysics Data System (ADS)

Ding, Peng; Zhang, Ye; Deng, Wei-Jian; Jia, Ping; Kuijper, Arjan

2018-07-01

Detection of objects from satellite optical remote sensing images is very important for many commercial and governmental applications. With the development of deep convolutional neural networks (deep CNNs), the field of object detection has seen tremendous advances. Currently, objects in satellite remote sensing images can be detected using deep CNNs. In general, optical remote sensing images contain many dense and small objects, and the use of the original Faster Regional CNN framework does not yield a suitably high precision. Therefore, after careful analysis we adopt dense convoluted networks, a multi-scale representation and various combinations of improvement schemes to enhance the structure of the base VGG16-Net for improving the precision. We propose an approach to reduce the test-time (detection time) and memory requirements. To validate the effectiveness of our approach, we perform experiments using satellite remote sensing image datasets of aircraft and automobiles. The results show that the improved network structure can detect objects in satellite optical remote sensing images more accurately and efficiently.

DeepFix: A Fully Convolutional Neural Network for Predicting Human Eye Fixations.

PubMed

Kruthiventi, Srinivas S S; Ayush, Kumar; Babu, R Venkatesh

2017-09-01

Understanding and predicting the human visual attention mechanism is an active area of research in the fields of neuroscience and computer vision. In this paper, we propose DeepFix, a fully convolutional neural network, which models the bottom-up mechanism of visual attention via saliency prediction. Unlike classical works, which characterize the saliency map using various hand-crafted features, our model automatically learns features in a hierarchical fashion and predicts the saliency map in an end-to-end manner. DeepFix is designed to capture semantics at multiple scales while taking global context into account, by using network layers with very large receptive fields. Generally, fully convolutional nets are spatially invariant-this prevents them from modeling location-dependent patterns (e.g., centre-bias). Our network handles this by incorporating a novel location-biased convolutional layer. We evaluate our model on multiple challenging saliency data sets and show that it achieves the state-of-the-art results.

Image inpainting and super-resolution using non-local recursive deep convolutional network with skip connections

NASA Astrophysics Data System (ADS)

Liu, Miaofeng

2017-07-01

In recent years, deep convolutional neural networks come into use in image inpainting and super-resolution in many fields. Distinct to most of the former methods requiring to know beforehand the local information for corrupted pixels, we propose a 20-depth fully convolutional network to learn an end-to-end mapping a dataset of damaged/ground truth subimage pairs realizing non-local blind inpainting and super-resolution. As there often exist image with huge corruptions or inpainting on a low-resolution image that the existing approaches unable to perform well, we also share parameters in local area of layers to achieve spatial recursion and enlarge the receptive field. To avoid the difficulty of training this deep neural network, skip-connections between symmetric convolutional layers are designed. Experimental results shows that the proposed method outperforms state-of-the-art methods for diverse corrupting and low-resolution conditions, it works excellently when realizing super-resolution and image inpainting simultaneously

Robustness of spiking Deep Belief Networks to noise and reduced bit precision of neuro-inspired hardware platforms.

PubMed

Stromatias, Evangelos; Neil, Daniel; Pfeiffer, Michael; Galluppi, Francesco; Furber, Steve B; Liu, Shih-Chii

2015-01-01

Increasingly large deep learning architectures, such as Deep Belief Networks (DBNs) are the focus of current machine learning research and achieve state-of-the-art results in different domains. However, both training and execution of large-scale Deep Networks require vast computing resources, leading to high power requirements and communication overheads. The on-going work on design and construction of spike-based hardware platforms offers an alternative for running deep neural networks with significantly lower power consumption, but has to overcome hardware limitations in terms of noise and limited weight precision, as well as noise inherent in the sensor signal. This article investigates how such hardware constraints impact the performance of spiking neural network implementations of DBNs. In particular, the influence of limited bit precision during execution and training, and the impact of silicon mismatch in the synaptic weight parameters of custom hybrid VLSI implementations is studied. Furthermore, the network performance of spiking DBNs is characterized with regard to noise in the spiking input signal. Our results demonstrate that spiking DBNs can tolerate very low levels of hardware bit precision down to almost two bits, and show that their performance can be improved by at least 30% through an adapted training mechanism that takes the bit precision of the target platform into account. Spiking DBNs thus present an important use-case for large-scale hybrid analog-digital or digital neuromorphic platforms such as SpiNNaker, which can execute large but precision-constrained deep networks in real time.

Robustness of spiking Deep Belief Networks to noise and reduced bit precision of neuro-inspired hardware platforms

PubMed Central

Stromatias, Evangelos; Neil, Daniel; Pfeiffer, Michael; Galluppi, Francesco; Furber, Steve B.; Liu, Shih-Chii

2015-01-01

Increasingly large deep learning architectures, such as Deep Belief Networks (DBNs) are the focus of current machine learning research and achieve state-of-the-art results in different domains. However, both training and execution of large-scale Deep Networks require vast computing resources, leading to high power requirements and communication overheads. The on-going work on design and construction of spike-based hardware platforms offers an alternative for running deep neural networks with significantly lower power consumption, but has to overcome hardware limitations in terms of noise and limited weight precision, as well as noise inherent in the sensor signal. This article investigates how such hardware constraints impact the performance of spiking neural network implementations of DBNs. In particular, the influence of limited bit precision during execution and training, and the impact of silicon mismatch in the synaptic weight parameters of custom hybrid VLSI implementations is studied. Furthermore, the network performance of spiking DBNs is characterized with regard to noise in the spiking input signal. Our results demonstrate that spiking DBNs can tolerate very low levels of hardware bit precision down to almost two bits, and show that their performance can be improved by at least 30% through an adapted training mechanism that takes the bit precision of the target platform into account. Spiking DBNs thus present an important use-case for large-scale hybrid analog-digital or digital neuromorphic platforms such as SpiNNaker, which can execute large but precision-constrained deep networks in real time. PMID:26217169

Using deep recurrent neural network for direct beam solar irradiance cloud screening

NASA Astrophysics Data System (ADS)

Chen, Maosi; Davis, John M.; Liu, Chaoshun; Sun, Zhibin; Zempila, Melina Maria; Gao, Wei

2017-09-01

Cloud screening is an essential procedure for in-situ calibration and atmospheric properties retrieval on (UV-)MultiFilter Rotating Shadowband Radiometer [(UV-)MFRSR]. Previous study has explored a cloud screening algorithm for direct-beam (UV-)MFRSR voltage measurements based on the stability assumption on a long time period (typically a half day or a whole day). To design such an algorithm requires in-depth understanding of radiative transfer and delicate data manipulation. Recent rapid developments on deep neural network and computation hardware have opened a window for modeling complicated End-to-End systems with a standardized strategy. In this study, a multi-layer dynamic bidirectional recurrent neural network is built for determining the cloudiness on each time point with a 17-year training dataset and tested with another 1-year dataset. The dataset is the daily 3-minute cosine corrected voltages, airmasses, and the corresponding cloud/clear-sky labels at two stations of the USDA UV-B Monitoring and Research Program. The results show that the optimized neural network model (3-layer, 250 hidden units, and 80 epochs of training) has an overall test accuracy of 97.87% (97.56% for the Oklahoma site and 98.16% for the Hawaii site). Generally, the neural network model grasps the key concept of the original model to use data in the entire day rather than short nearby measurements to perform cloud screening. A scrutiny of the logits layer suggests that the neural network model automatically learns a way to calculate a quantity similar to total optical depth and finds an appropriate threshold for cloud screening.

Visual Saliency Detection Based on Multiscale Deep CNN Features.

PubMed

Guanbin Li; Yizhou Yu

2016-11-01

Visual saliency is a fundamental problem in both cognitive and computational sciences, including computer vision. In this paper, we discover that a high-quality visual saliency model can be learned from multiscale features extracted using deep convolutional neural networks (CNNs), which have had many successes in visual recognition tasks. For learning such saliency models, we introduce a neural network architecture, which has fully connected layers on top of CNNs responsible for feature extraction at three different scales. The penultimate layer of our neural network has been confirmed to be a discriminative high-level feature vector for saliency detection, which we call deep contrast feature. To generate a more robust feature, we integrate handcrafted low-level features with our deep contrast feature. To promote further research and evaluation of visual saliency models, we also construct a new large database of 4447 challenging images and their pixelwise saliency annotations. Experimental results demonstrate that our proposed method is capable of achieving the state-of-the-art performance on all public benchmarks, improving the F-measure by 6.12% and 10%, respectively, on the DUT-OMRON data set and our new data set (HKU-IS), and lowering the mean absolute error by 9% and 35.3%, respectively, on these two data sets.

A Deep Learning-Based Method for Similar Patient Question Retrieval in Chinese.

PubMed

Tang, Guo Yu; Ni, Yuan; Xie, Guo Tong; Fan, Xin Li; Shi, Yan Ling

2017-01-01

The online patient question and answering (Q&A) system, either as a website or a mobile application, attracts an increasing number of users in China. Patients will post their questions and the registered doctors then provide the corresponding answers. A large amount of questions with answers from doctors are accumulated. Instead of awaiting the response from a doctor, the newly posted question could be quickly answered by finding a semantically equivalent question from the Q&A achive. In this study, we investigated a novel deep learning based method to retrieve the similar patient question in Chinese. An unsupervised learning algorithm using deep neural network is performed on the corpus to generate the word embedding. The word embedding was then used as the input to a supervised learning algorithm using a designed deep neural network, i.e. the supervised neural attention model (SNA), to predict the similarity between two questions. The experimental results showed that our SNA method achieved P@1 = 77% and P@5 = 84%, which outperformed all other compared methods.

Deep Logic Networks: Inserting and Extracting Knowledge From Deep Belief Networks.

PubMed

Tran, Son N; d'Avila Garcez, Artur S

2018-02-01

Developments in deep learning have seen the use of layerwise unsupervised learning combined with supervised learning for fine-tuning. With this layerwise approach, a deep network can be seen as a more modular system that lends itself well to learning representations. In this paper, we investigate whether such modularity can be useful to the insertion of background knowledge into deep networks, whether it can improve learning performance when it is available, and to the extraction of knowledge from trained deep networks, and whether it can offer a better understanding of the representations learned by such networks. To this end, we use a simple symbolic language-a set of logical rules that we call confidence rules-and show that it is suitable for the representation of quantitative reasoning in deep networks. We show by knowledge extraction that confidence rules can offer a low-cost representation for layerwise networks (or restricted Boltzmann machines). We also show that layerwise extraction can produce an improvement in the accuracy of deep belief networks. Furthermore, the proposed symbolic characterization of deep networks provides a novel method for the insertion of prior knowledge and training of deep networks. With the use of this method, a deep neural-symbolic system is proposed and evaluated, with the experimental results indicating that modularity through the use of confidence rules and knowledge insertion can be beneficial to network performance.

Nonlinear analysis and synthesis of video images using deep dynamic bottleneck neural networks for face recognition.

PubMed

Moghadam, Saeed Montazeri; Seyyedsalehi, Seyyed Ali

2018-05-31

Nonlinear components extracted from deep structures of bottleneck neural networks exhibit a great ability to express input space in a low-dimensional manifold. Sharing and combining the components boost the capability of the neural networks to synthesize and interpolate new and imaginary data. This synthesis is possibly a simple model of imaginations in human brain where the components are expressed in a nonlinear low dimensional manifold. The current paper introduces a novel Dynamic Deep Bottleneck Neural Network to analyze and extract three main features of videos regarding the expression of emotions on the face. These main features are identity, emotion and expression intensity that are laid in three different sub-manifolds of one nonlinear general manifold. The proposed model enjoying the advantages of recurrent networks was used to analyze the sequence and dynamics of information in videos. It is noteworthy to mention that this model also has also the potential to synthesize new videos showing variations of one specific emotion on the face of unknown subjects. Experiments on discrimination and recognition ability of extracted components showed that the proposed model has an average of 97.77% accuracy in recognition of six prominent emotions (Fear, Surprise, Sadness, Anger, Disgust, and Happiness), and 78.17% accuracy in the recognition of intensity. The produced videos revealed variations from neutral to the apex of an emotion on the face of the unfamiliar test subject which is on average 0.8 similar to reference videos in the scale of the SSIM method. Copyright © 2018 Elsevier Ltd. All rights reserved.

Imbalance aware lithography hotspot detection: a deep learning approach

NASA Astrophysics Data System (ADS)

Yang, Haoyu; Luo, Luyang; Su, Jing; Lin, Chenxi; Yu, Bei

2017-07-01

With the advancement of very large scale integrated circuits (VLSI) technology nodes, lithographic hotspots become a serious problem that affects manufacture yield. Lithography hotspot detection at the post-OPC stage is imperative to check potential circuit failures when transferring designed patterns onto silicon wafers. Although conventional lithography hotspot detection methods, such as machine learning, have gained satisfactory performance, with the extreme scaling of transistor feature size and layout patterns growing in complexity, conventional methodologies may suffer from performance degradation. For example, manual or ad hoc feature extraction in a machine learning framework may lose important information when predicting potential errors in ultra-large-scale integrated circuit masks. We present a deep convolutional neural network (CNN) that targets representative feature learning in lithography hotspot detection. We carefully analyze the impact and effectiveness of different CNN hyperparameters, through which a hotspot-detection-oriented neural network model is established. Because hotspot patterns are always in the minority in VLSI mask design, the training dataset is highly imbalanced. In this situation, a neural network is no longer reliable, because a trained model with high classification accuracy may still suffer from a high number of false negative results (missing hotspots), which is fatal in hotspot detection problems. To address the imbalance problem, we further apply hotspot upsampling and random-mirror flipping before training the network. Experimental results show that our proposed neural network model achieves comparable or better performance on the ICCAD 2012 contest benchmark compared to state-of-the-art hotspot detectors based on deep or representative machine leaning.

Deep neural networks show an equivalent and often superior performance to dermatologists in onychomycosis diagnosis: Automatic construction of onychomycosis datasets by region-based convolutional deep neural network.

PubMed

Han, Seung Seog; Park, Gyeong Hun; Lim, Woohyung; Kim, Myoung Shin; Na, Jung Im; Park, Ilwoo; Chang, Sung Eun

2018-01-01

Although there have been reports of the successful diagnosis of skin disorders using deep learning, unrealistically large clinical image datasets are required for artificial intelligence (AI) training. We created datasets of standardized nail images using a region-based convolutional neural network (R-CNN) trained to distinguish the nail from the background. We used R-CNN to generate training datasets of 49,567 images, which we then used to fine-tune the ResNet-152 and VGG-19 models. The validation datasets comprised 100 and 194 images from Inje University (B1 and B2 datasets, respectively), 125 images from Hallym University (C dataset), and 939 images from Seoul National University (D dataset). The AI (ensemble model; ResNet-152 + VGG-19 + feedforward neural networks) results showed test sensitivity/specificity/ area under the curve values of (96.0 / 94.7 / 0.98), (82.7 / 96.7 / 0.95), (92.3 / 79.3 / 0.93), (87.7 / 69.3 / 0.82) for the B1, B2, C, and D datasets. With a combination of the B1 and C datasets, the AI Youden index was significantly (p = 0.01) higher than that of 42 dermatologists doing the same assessment manually. For B1+C and B2+ D dataset combinations, almost none of the dermatologists performed as well as the AI. By training with a dataset comprising 49,567 images, we achieved a diagnostic accuracy for onychomycosis using deep learning that was superior to that of most of the dermatologists who participated in this study.

Image quality assessment using deep convolutional networks

NASA Astrophysics Data System (ADS)

Li, Yezhou; Ye, Xiang; Li, Yong

2017-12-01

This paper proposes a method of accurately assessing image quality without a reference image by using a deep convolutional neural network. Existing training based methods usually utilize a compact set of linear filters for learning features of images captured by different sensors to assess their quality. These methods may not be able to learn the semantic features that are intimately related with the features used in human subject assessment. Observing this drawback, this work proposes training a deep convolutional neural network (CNN) with labelled images for image quality assessment. The ReLU in the CNN allows non-linear transformations for extracting high-level image features, providing a more reliable assessment of image quality than linear filters. To enable the neural network to take images of any arbitrary size as input, the spatial pyramid pooling (SPP) is introduced connecting the top convolutional layer and the fully-connected layer. In addition, the SPP makes the CNN robust to object deformations to a certain extent. The proposed method taking an image as input carries out an end-to-end learning process, and outputs the quality of the image. It is tested on public datasets. Experimental results show that it outperforms existing methods by a large margin and can accurately assess the image quality on images taken by different sensors of varying sizes.

Target recognition based on convolutional neural network

NASA Astrophysics Data System (ADS)

Wang, Liqiang; Wang, Xin; Xi, Fubiao; Dong, Jian

2017-11-01

One of the important part of object target recognition is the feature extraction, which can be classified into feature extraction and automatic feature extraction. The traditional neural network is one of the automatic feature extraction methods, while it causes high possibility of over-fitting due to the global connection. The deep learning algorithm used in this paper is a hierarchical automatic feature extraction method, trained with the layer-by-layer convolutional neural network (CNN), which can extract the features from lower layers to higher layers. The features are more discriminative and it is beneficial to the object target recognition.

TernaryNet: faster deep model inference without GPUs for medical 3D segmentation using sparse and binary convolutions.

PubMed

Heinrich, Mattias P; Blendowski, Max; Oktay, Ozan

2018-05-30

Deep convolutional neural networks (DCNN) are currently ubiquitous in medical imaging. While their versatility and high-quality results for common image analysis tasks including segmentation, localisation and prediction is astonishing, the large representational power comes at the cost of highly demanding computational effort. This limits their practical applications for image-guided interventions and diagnostic (point-of-care) support using mobile devices without graphics processing units (GPU). We propose a new scheme that approximates both trainable weights and neural activations in deep networks by ternary values and tackles the open question of backpropagation when dealing with non-differentiable functions. Our solution enables the removal of the expensive floating-point matrix multiplications throughout any convolutional neural network and replaces them by energy- and time-preserving binary operators and population counts. We evaluate our approach for the segmentation of the pancreas in CT. Here, our ternary approximation within a fully convolutional network leads to more than 90% memory reductions and high accuracy (without any post-processing) with a Dice overlap of 71.0% that comes close to the one obtained when using networks with high-precision weights and activations. We further provide a concept for sub-second inference without GPUs and demonstrate significant improvements in comparison with binary quantisation and without our proposed ternary hyperbolic tangent continuation. We present a key enabling technique for highly efficient DCNN inference without GPUs that will help to bring the advances of deep learning to practical clinical applications. It has also great promise for improving accuracies in large-scale medical data retrieval.

Neural Decoder for Topological Codes

NASA Astrophysics Data System (ADS)

Torlai, Giacomo; Melko, Roger G.

2017-07-01

We present an algorithm for error correction in topological codes that exploits modern machine learning techniques. Our decoder is constructed from a stochastic neural network called a Boltzmann machine, of the type extensively used in deep learning. We provide a general prescription for the training of the network and a decoding strategy that is applicable to a wide variety of stabilizer codes with very little specialization. We demonstrate the neural decoder numerically on the well-known two-dimensional toric code with phase-flip errors.

Deep Learning in Medical Imaging: General Overview

PubMed Central

Lee, June-Goo; Jun, Sanghoon; Cho, Young-Won; Lee, Hyunna; Kim, Guk Bae

2017-01-01

The artificial neural network (ANN)–a machine learning technique inspired by the human neuronal synapse system–was introduced in the 1950s. However, the ANN was previously limited in its ability to solve actual problems, due to the vanishing gradient and overfitting problems with training of deep architecture, lack of computing power, and primarily the absence of sufficient data to train the computer system. Interest in this concept has lately resurfaced, due to the availability of big data, enhanced computing power with the current graphics processing units, and novel algorithms to train the deep neural network. Recent studies on this technology suggest its potentially to perform better than humans in some visual and auditory recognition tasks, which may portend its applications in medicine and healthcare, especially in medical imaging, in the foreseeable future. This review article offers perspectives on the history, development, and applications of deep learning technology, particularly regarding its applications in medical imaging. PMID:28670152

Deep Learning in Medical Imaging: General Overview.

PubMed

Lee, June-Goo; Jun, Sanghoon; Cho, Young-Won; Lee, Hyunna; Kim, Guk Bae; Seo, Joon Beom; Kim, Namkug

2017-01-01

The artificial neural network (ANN)-a machine learning technique inspired by the human neuronal synapse system-was introduced in the 1950s. However, the ANN was previously limited in its ability to solve actual problems, due to the vanishing gradient and overfitting problems with training of deep architecture, lack of computing power, and primarily the absence of sufficient data to train the computer system. Interest in this concept has lately resurfaced, due to the availability of big data, enhanced computing power with the current graphics processing units, and novel algorithms to train the deep neural network. Recent studies on this technology suggest its potentially to perform better than humans in some visual and auditory recognition tasks, which may portend its applications in medicine and healthcare, especially in medical imaging, in the foreseeable future. This review article offers perspectives on the history, development, and applications of deep learning technology, particularly regarding its applications in medical imaging.

Effects of Electrical and Optogenetic Deep Brain Stimulation on Synchronized Oscillatory Activity in Parkinsonian Basal Ganglia.

PubMed

Ratnadurai-Giridharan, Shivakeshavan; Cheung, Chung C; Rubchinsky, Leonid L

2017-11-01

Conventional deep brain stimulation of basal ganglia uses high-frequency regular electrical pulses to treat Parkinsonian motor symptoms but has a series of limitations. Relatively new and not yet clinically tested, optogenetic stimulation is an effective experimental stimulation technique to affect pathological network dynamics. We compared the effects of electrical and optogenetic stimulation of the basal gangliaon the pathologicalParkinsonian rhythmic neural activity. We studied the network response to electrical stimulation and excitatory and inhibitory optogenetic stimulations. Different stimulations exhibit different interactions with pathological activity in the network. We studied these interactions for different network and stimulation parameter values. Optogenetic stimulation was found to be more efficient than electrical stimulation in suppressing pathological rhythmicity. Our findings indicate that optogenetic control of neural synchrony may be more efficacious than electrical control because of the different ways of how stimulations interact with network dynamics.

Deep neural network-based domain adaptation for classification of remote sensing images

NASA Astrophysics Data System (ADS)

Ma, Li; Song, Jiazhen

2017-10-01

We investigate the effectiveness of deep neural network for cross-domain classification of remote sensing images in this paper. In the network, class centroid alignment is utilized as a domain adaptation strategy, making the network able to transfer knowledge from the source domain to target domain on a per-class basis. Since predicted labels of target data should be used to estimate the centroid of each class, we use overall centroid alignment as a coarse domain adaptation method to improve the estimation accuracy. In addition, rectified linear unit is used as the activation function to produce sparse features, which may improve the separation capability. The proposed network can provide both aligned features and an adaptive classifier, as well as obtain label-free classification of target domain data. The experimental results using Hyperion, NCALM, and WorldView-2 remote sensing images demonstrated the effectiveness of the proposed approach.

Mastering the game of Go with deep neural networks and tree search.

PubMed

Silver, David; Huang, Aja; Maddison, Chris J; Guez, Arthur; Sifre, Laurent; van den Driessche, George; Schrittwieser, Julian; Antonoglou, Ioannis; Panneershelvam, Veda; Lanctot, Marc; Dieleman, Sander; Grewe, Dominik; Nham, John; Kalchbrenner, Nal; Sutskever, Ilya; Lillicrap, Timothy; Leach, Madeleine; Kavukcuoglu, Koray; Graepel, Thore; Hassabis, Demis

2016-01-28

The game of Go has long been viewed as the most challenging of classic games for artificial intelligence owing to its enormous search space and the difficulty of evaluating board positions and moves. Here we introduce a new approach to computer Go that uses 'value networks' to evaluate board positions and 'policy networks' to select moves. These deep neural networks are trained by a novel combination of supervised learning from human expert games, and reinforcement learning from games of self-play. Without any lookahead search, the neural networks play Go at the level of state-of-the-art Monte Carlo tree search programs that simulate thousands of random games of self-play. We also introduce a new search algorithm that combines Monte Carlo simulation with value and policy networks. Using this search algorithm, our program AlphaGo achieved a 99.8% winning rate against other Go programs, and defeated the human European Go champion by 5 games to 0. This is the first time that a computer program has defeated a human professional player in the full-sized game of Go, a feat previously thought to be at least a decade away.

Event Recognition Based on Deep Learning in Chinese Texts

PubMed Central

Zhang, Yajun; Liu, Zongtian; Zhou, Wen

2016-01-01

Event recognition is the most fundamental and critical task in event-based natural language processing systems. Existing event recognition methods based on rules and shallow neural networks have certain limitations. For example, extracting features using methods based on rules is difficult; methods based on shallow neural networks converge too quickly to a local minimum, resulting in low recognition precision. To address these problems, we propose the Chinese emergency event recognition model based on deep learning (CEERM). Firstly, we use a word segmentation system to segment sentences. According to event elements labeled in the CEC 2.0 corpus, we classify words into five categories: trigger words, participants, objects, time and location. Each word is vectorized according to the following six feature layers: part of speech, dependency grammar, length, location, distance between trigger word and core word and trigger word frequency. We obtain deep semantic features of words by training a feature vector set using a deep belief network (DBN), then analyze those features in order to identify trigger words by means of a back propagation neural network. Extensive testing shows that the CEERM achieves excellent recognition performance, with a maximum F-measure value of 85.17%. Moreover, we propose the dynamic-supervised DBN, which adds supervised fine-tuning to a restricted Boltzmann machine layer by monitoring its training performance. Test analysis reveals that the new DBN improves recognition performance and effectively controls the training time. Although the F-measure increases to 88.11%, the training time increases by only 25.35%. PMID:27501231

Event Recognition Based on Deep Learning in Chinese Texts.

PubMed

Zhang, Yajun; Liu, Zongtian; Zhou, Wen

2016-01-01

Event recognition is the most fundamental and critical task in event-based natural language processing systems. Existing event recognition methods based on rules and shallow neural networks have certain limitations. For example, extracting features using methods based on rules is difficult; methods based on shallow neural networks converge too quickly to a local minimum, resulting in low recognition precision. To address these problems, we propose the Chinese emergency event recognition model based on deep learning (CEERM). Firstly, we use a word segmentation system to segment sentences. According to event elements labeled in the CEC 2.0 corpus, we classify words into five categories: trigger words, participants, objects, time and location. Each word is vectorized according to the following six feature layers: part of speech, dependency grammar, length, location, distance between trigger word and core word and trigger word frequency. We obtain deep semantic features of words by training a feature vector set using a deep belief network (DBN), then analyze those features in order to identify trigger words by means of a back propagation neural network. Extensive testing shows that the CEERM achieves excellent recognition performance, with a maximum F-measure value of 85.17%. Moreover, we propose the dynamic-supervised DBN, which adds supervised fine-tuning to a restricted Boltzmann machine layer by monitoring its training performance. Test analysis reveals that the new DBN improves recognition performance and effectively controls the training time. Although the F-measure increases to 88.11%, the training time increases by only 25.35%.

Deep learning for medical image segmentation - using the IBM TrueNorth neurosynaptic system

NASA Astrophysics Data System (ADS)

Moran, Steven; Gaonkar, Bilwaj; Whitehead, William; Wolk, Aidan; Macyszyn, Luke; Iyer, Subramanian S.

2018-03-01

Deep convolutional neural networks have found success in semantic image segmentation tasks in computer vision and medical imaging. These algorithms are executed on conventional von Neumann processor architectures or GPUs. This is suboptimal. Neuromorphic processors that replicate the structure of the brain are better-suited to train and execute deep learning models for image segmentation by relying on massively-parallel processing. However, given that they closely emulate the human brain, on-chip hardware and digital memory limitations also constrain them. Adapting deep learning models to execute image segmentation tasks on such chips, requires specialized training and validation. In this work, we demonstrate for the first-time, spinal image segmentation performed using a deep learning network implemented on neuromorphic hardware of the IBM TrueNorth Neurosynaptic System and validate the performance of our network by comparing it to human-generated segmentations of spinal vertebrae and disks. To achieve this on neuromorphic hardware, the training model constrains the coefficients of individual neurons to {-1,0,1} using the Energy Efficient Deep Neuromorphic (EEDN)1 networks training algorithm. Given the 1 million neurons and 256 million synapses, the scale and size of the neural network implemented by the IBM TrueNorth allows us to execute the requisite mapping between segmented images and non-uniform intensity MR images >20 times faster than on a GPU-accelerated network and using <0.1 W. This speed and efficiency implies that a trained neuromorphic chip can be deployed in intra-operative environments where real-time medical image segmentation is necessary.

Deep Recurrent Neural Networks for Supernovae Classification

NASA Astrophysics Data System (ADS)

Charnock, Tom; Moss, Adam

2017-03-01

We apply deep recurrent neural networks, which are capable of learning complex sequential information, to classify supernovae (code available at https://github.com/adammoss/supernovae). The observational time and filter fluxes are used as inputs to the network, but since the inputs are agnostic, additional data such as host galaxy information can also be included. Using the Supernovae Photometric Classification Challenge (SPCC) data, we find that deep networks are capable of learning about light curves, however the performance of the network is highly sensitive to the amount of training data. For a training size of 50% of the representational SPCC data set (around 104 supernovae) we obtain a type-Ia versus non-type-Ia classification accuracy of 94.7%, an area under the Receiver Operating Characteristic curve AUC of 0.986 and an SPCC figure-of-merit F 1 = 0.64. When using only the data for the early-epoch challenge defined by the SPCC, we achieve a classification accuracy of 93.1%, AUC of 0.977, and F 1 = 0.58, results almost as good as with the whole light curve. By employing bidirectional neural networks, we can acquire impressive classification results between supernovae types I, II and III at an accuracy of 90.4% and AUC of 0.974. We also apply a pre-trained model to obtain classification probabilities as a function of time and show that it can give early indications of supernovae type. Our method is competitive with existing algorithms and has applications for future large-scale photometric surveys.

ACTIVIS: Visual Exploration of Industry-Scale Deep Neural Network Models.

PubMed

Kahng, Minsuk; Andrews, Pierre Y; Kalro, Aditya; Polo Chau, Duen Horng

2017-08-30

While deep learning models have achieved state-of-the-art accuracies for many prediction tasks, understanding these models remains a challenge. Despite the recent interest in developing visual tools to help users interpret deep learning models, the complexity and wide variety of models deployed in industry, and the large-scale datasets that they used, pose unique design challenges that are inadequately addressed by existing work. Through participatory design sessions with over 15 researchers and engineers at Facebook, we have developed, deployed, and iteratively improved ACTIVIS, an interactive visualization system for interpreting large-scale deep learning models and results. By tightly integrating multiple coordinated views, such as a computation graph overview of the model architecture, and a neuron activation view for pattern discovery and comparison, users can explore complex deep neural network models at both the instance- and subset-level. ACTIVIS has been deployed on Facebook's machine learning platform. We present case studies with Facebook researchers and engineers, and usage scenarios of how ACTIVIS may work with different models.

Comment on ‘Deep convolutional neural network with transfer learning for rectum toxicity prediction in cervical cancer radiotherapy: a feasibility study’

NASA Astrophysics Data System (ADS)

Valdes, Gilmer; Interian, Yannet

2018-03-01

The application of machine learning (ML) presents tremendous opportunities for the field of oncology, thus we read ‘Deep convolutional neural network with transfer learning for rectum toxicity prediction in cervical cancer radiotherapy: a feasibility study’ with great interest. In this article, the authors used state of the art techniques: a pre-trained convolutional neural network (VGG-16 CNN), transfer learning, data augmentation, drop out and early stopping, all of which are directly responsible for the success and the excitement that these algorithms have created in other fields. We believe that the use of these techniques can offer tremendous opportunities in the field of Medical Physics and as such we would like to praise the authors for their pioneering application to the field of Radiation Oncology. That being said, given that the field of Medical Physics has unique characteristics that differentiate us from those fields where these techniques have been applied successfully, we would like to raise some points for future discussion and follow up studies that could help the community understand the limitations and nuances of deep learning techniques.

Automated Depression Analysis Using Convolutional Neural Networks from Speech.

PubMed

He, Lang; Cao, Cui

2018-05-28

To help clinicians to efficiently diagnose the severity of a person's depression, the affective computing community and the artificial intelligence field have shown a growing interest in designing automated systems. The speech features have useful information for the diagnosis of depression. However, manually designing and domain knowledge are still important for the selection of the feature, which makes the process labor consuming and subjective. In recent years, deep-learned features based on neural networks have shown superior performance to hand-crafted features in various areas. In this paper, to overcome the difficulties mentioned above, we propose a combination of hand-crafted and deep-learned features which can effectively measure the severity of depression from speech. In the proposed method, Deep Convolutional Neural Networks (DCNN) are firstly built to learn deep-learned features from spectrograms and raw speech waveforms. Then we manually extract the state-of-the-art texture descriptors named median robust extended local binary patterns (MRELBP) from spectrograms. To capture the complementary information within the hand-crafted features and deep-learned features, we propose joint fine-tuning layers to combine the raw and spectrogram DCNN to boost the depression recognition performance. Moreover, to address the problems with small samples, a data augmentation method was proposed. Experiments conducted on AVEC2013 and AVEC2014 depression databases show that our approach is robust and effective for the diagnosis of depression when compared to state-of-the-art audio-based methods. Copyright © 2018. Published by Elsevier Inc.

Comparative Studies of Prediction Strategies for Solar X-ray Time Series

NASA Astrophysics Data System (ADS)

Muranushi, T.; Hattori, T.; Jin, Q.; Hishinuma, T.; Tominaga, M.; Nakagawa, K.; Fujiwara, Y.; Nakamura, T.; Sakaue, T.; Takahashi, T.; Seki, D.; Namekata, K.; Tei, A.; Ban, M.; Kawamura, A. D.; Hada-Muranushi, Y.; Asai, A.; Nemoto, S.; Shibata, K.

2016-12-01

Crucial virtues for operational space weather forecast are real-timeforecast ability, forecast precision and customizability to userneeds. The recent development of deep-learning makes it veryattractive to space weather, because (1) it learns gradually incomingdata, (2) it exhibits superior accuracy over conventional algorithmsin many fields, and (3) it makes the customization of the forecasteasier because it accepts raw images.However, the best deep-learning applications are only attainable bycareful human designers that understands both the mechanism of deeplearning and the application field. Therefore, we need to foster youngresearchers to enter the field of machine-learning aided forecast. So,we have held a seminar every Monday with undergraduate and graduatestudents from May to August 2016.We will review the current status of space weather science and theautomated real-time space weather forecast engine UFCORIN. Then, weintroduce the deep-learning space weather forecast environments wehave set up using Python and Chainer on students' laptop computers.We have started from simple image classification neural network, thenimplemented space-weather neural network that predicts future X-rayflux of the Sun based on the past X-ray lightcurve and magnetic fieldline-of-sight images.In order to perform each forecast faster, we have focused on simplelightcurve-to-lightcurve forecast, and performed comparative surveysby changing following parameters: The size and topology of the neural network Batchsize Neural network hyperparameters such as learning rates to optimize the preduction accuracy, and time for prediction.We have found how to design compact, fast but accurate neural networkto perform forecast. Our forecasters can perform predictionexperiment for four-year timespan in a few minutes, and achieveslog-scale errors of the order of 1. Our studies is ongoing, and inour talk we will review our progress till December.

DeepNAT: Deep convolutional neural network for segmenting neuroanatomy.

PubMed

Wachinger, Christian; Reuter, Martin; Klein, Tassilo

2018-04-15

We introduce DeepNAT, a 3D Deep convolutional neural network for the automatic segmentation of NeuroAnaTomy in T1-weighted magnetic resonance images. DeepNAT is an end-to-end learning-based approach to brain segmentation that jointly learns an abstract feature representation and a multi-class classification. We propose a 3D patch-based approach, where we do not only predict the center voxel of the patch but also neighbors, which is formulated as multi-task learning. To address a class imbalance problem, we arrange two networks hierarchically, where the first one separates foreground from background, and the second one identifies 25 brain structures on the foreground. Since patches lack spatial context, we augment them with coordinates. To this end, we introduce a novel intrinsic parameterization of the brain volume, formed by eigenfunctions of the Laplace-Beltrami operator. As network architecture, we use three convolutional layers with pooling, batch normalization, and non-linearities, followed by fully connected layers with dropout. The final segmentation is inferred from the probabilistic output of the network with a 3D fully connected conditional random field, which ensures label agreement between close voxels. The roughly 2.7million parameters in the network are learned with stochastic gradient descent. Our results show that DeepNAT compares favorably to state-of-the-art methods. Finally, the purely learning-based method may have a high potential for the adaptation to young, old, or diseased brains by fine-tuning the pre-trained network with a small training sample on the target application, where the availability of larger datasets with manual annotations may boost the overall segmentation accuracy in the future. Copyright © 2017 Elsevier Inc. All rights reserved.

Deep Neural Networks Based Recognition of Plant Diseases by Leaf Image Classification.

PubMed

Sladojevic, Srdjan; Arsenovic, Marko; Anderla, Andras; Culibrk, Dubravko; Stefanovic, Darko

2016-01-01

The latest generation of convolutional neural networks (CNNs) has achieved impressive results in the field of image classification. This paper is concerned with a new approach to the development of plant disease recognition model, based on leaf image classification, by the use of deep convolutional networks. Novel way of training and the methodology used facilitate a quick and easy system implementation in practice. The developed model is able to recognize 13 different types of plant diseases out of healthy leaves, with the ability to distinguish plant leaves from their surroundings. According to our knowledge, this method for plant disease recognition has been proposed for the first time. All essential steps required for implementing this disease recognition model are fully described throughout the paper, starting from gathering images in order to create a database, assessed by agricultural experts. Caffe, a deep learning framework developed by Berkley Vision and Learning Centre, was used to perform the deep CNN training. The experimental results on the developed model achieved precision between 91% and 98%, for separate class tests, on average 96.3%.

Deep Neural Networks Based Recognition of Plant Diseases by Leaf Image Classification

PubMed Central

Sladojevic, Srdjan; Arsenovic, Marko; Culibrk, Dubravko; Stefanovic, Darko

2016-01-01

The latest generation of convolutional neural networks (CNNs) has achieved impressive results in the field of image classification. This paper is concerned with a new approach to the development of plant disease recognition model, based on leaf image classification, by the use of deep convolutional networks. Novel way of training and the methodology used facilitate a quick and easy system implementation in practice. The developed model is able to recognize 13 different types of plant diseases out of healthy leaves, with the ability to distinguish plant leaves from their surroundings. According to our knowledge, this method for plant disease recognition has been proposed for the first time. All essential steps required for implementing this disease recognition model are fully described throughout the paper, starting from gathering images in order to create a database, assessed by agricultural experts. Caffe, a deep learning framework developed by Berkley Vision and Learning Centre, was used to perform the deep CNN training. The experimental results on the developed model achieved precision between 91% and 98%, for separate class tests, on average 96.3%. PMID:27418923

DeepMoon: Convolutional neural network trainer to identify moon craters

NASA Astrophysics Data System (ADS)

Silburt, Ari; Zhu, Chenchong; Ali-Dib, Mohamad; Menou, Kristen; Jackson, Alan

2018-05-01

DeepMoon trains a convolutional neural net using data derived from a global digital elevation map (DEM) and catalog of craters to recognize craters on the Moon. The TensorFlow-based pipeline code is divided into three parts. The first generates a set images of the Moon randomly cropped from the DEM, with corresponding crater positions and radii. The second trains a convnet using this data, and the third validates the convnet's predictions.

Particle swarm optimization-based automatic parameter selection for deep neural networks and its applications in large-scale and high-dimensional data

PubMed Central

2017-01-01

In this paper, we propose a new automatic hyperparameter selection approach for determining the optimal network configuration (network structure and hyperparameters) for deep neural networks using particle swarm optimization (PSO) in combination with a steepest gradient descent algorithm. In the proposed approach, network configurations were coded as a set of real-number m-dimensional vectors as the individuals of the PSO algorithm in the search procedure. During the search procedure, the PSO algorithm is employed to search for optimal network configurations via the particles moving in a finite search space, and the steepest gradient descent algorithm is used to train the DNN classifier with a few training epochs (to find a local optimal solution) during the population evaluation of PSO. After the optimization scheme, the steepest gradient descent algorithm is performed with more epochs and the final solutions (pbest and gbest) of the PSO algorithm to train a final ensemble model and individual DNN classifiers, respectively. The local search ability of the steepest gradient descent algorithm and the global search capabilities of the PSO algorithm are exploited to determine an optimal solution that is close to the global optimum. We constructed several experiments on hand-written characters and biological activity prediction datasets to show that the DNN classifiers trained by the network configurations expressed by the final solutions of the PSO algorithm, employed to construct an ensemble model and individual classifier, outperform the random approach in terms of the generalization performance. Therefore, the proposed approach can be regarded an alternative tool for automatic network structure and parameter selection for deep neural networks. PMID:29236718

Classify epithelium-stroma in histopathological images based on deep transferable network.

PubMed

Yu, X; Zheng, H; Liu, C; Huang, Y; Ding, X

2018-04-20

Recently, the deep learning methods have received more attention in histopathological image analysis. However, the traditional deep learning methods assume that training data and test data have the same distributions, which causes certain limitations in real-world histopathological applications. However, it is costly to recollect a large amount of labeled histology data to train a new neural network for each specified image acquisition procedure even for similar tasks. In this paper, an unsupervised domain adaptation is introduced into a typical deep convolutional neural network (CNN) model to mitigate the repeating of the labels. The unsupervised domain adaptation is implemented by adding two regularisation terms, namely the feature-based adaptation and entropy minimisation, to the object function of a widely used CNN model called the AlexNet. Three independent public epithelium-stroma datasets were used to verify the proposed method. The experimental results have demonstrated that in the epithelium-stroma classification, the proposed method can achieve better performance than the commonly used deep learning methods and some existing deep domain adaptation methods. Therefore, the proposed method can be considered as a better option for the real-world applications of histopathological image analysis because there is no requirement for recollection of large-scale labeled data for every specified domain. © 2018 The Authors Journal of Microscopy © 2018 Royal Microscopical Society.

Overview of deep learning in medical imaging.

PubMed

Suzuki, Kenji

2017-09-01

The use of machine learning (ML) has been increasing rapidly in the medical imaging field, including computer-aided diagnosis (CAD), radiomics, and medical image analysis. Recently, an ML area called deep learning emerged in the computer vision field and became very popular in many fields. It started from an event in late 2012, when a deep-learning approach based on a convolutional neural network (CNN) won an overwhelming victory in the best-known worldwide computer vision competition, ImageNet Classification. Since then, researchers in virtually all fields, including medical imaging, have started actively participating in the explosively growing field of deep learning. In this paper, the area of deep learning in medical imaging is overviewed, including (1) what was changed in machine learning before and after the introduction of deep learning, (2) what is the source of the power of deep learning, (3) two major deep-learning models: a massive-training artificial neural network (MTANN) and a convolutional neural network (CNN), (4) similarities and differences between the two models, and (5) their applications to medical imaging. This review shows that ML with feature input (or feature-based ML) was dominant before the introduction of deep learning, and that the major and essential difference between ML before and after deep learning is the learning of image data directly without object segmentation or feature extraction; thus, it is the source of the power of deep learning, although the depth of the model is an important attribute. The class of ML with image input (or image-based ML) including deep learning has a long history, but recently gained popularity due to the use of the new terminology, deep learning. There are two major models in this class of ML in medical imaging, MTANN and CNN, which have similarities as well as several differences. In our experience, MTANNs were substantially more efficient in their development, had a higher performance, and required a lesser number of training cases than did CNNs. "Deep learning", or ML with image input, in medical imaging is an explosively growing, promising field. It is expected that ML with image input will be the mainstream area in the field of medical imaging in the next few decades.

DeepFruits: A Fruit Detection System Using Deep Neural Networks

PubMed Central

Sa, Inkyu; Ge, Zongyuan; Dayoub, Feras; Upcroft, Ben; Perez, Tristan; McCool, Chris

2016-01-01

This paper presents a novel approach to fruit detection using deep convolutional neural networks. The aim is to build an accurate, fast and reliable fruit detection system, which is a vital element of an autonomous agricultural robotic platform; it is a key element for fruit yield estimation and automated harvesting. Recent work in deep neural networks has led to the development of a state-of-the-art object detector termed Faster Region-based CNN (Faster R-CNN). We adapt this model, through transfer learning, for the task of fruit detection using imagery obtained from two modalities: colour (RGB) and Near-Infrared (NIR). Early and late fusion methods are explored for combining the multi-modal (RGB and NIR) information. This leads to a novel multi-modal Faster R-CNN model, which achieves state-of-the-art results compared to prior work with the F1 score, which takes into account both precision and recall performances improving from 0.807 to 0.838 for the detection of sweet pepper. In addition to improved accuracy, this approach is also much quicker to deploy for new fruits, as it requires bounding box annotation rather than pixel-level annotation (annotating bounding boxes is approximately an order of magnitude quicker to perform). The model is retrained to perform the detection of seven fruits, with the entire process taking four hours to annotate and train the new model per fruit. PMID:27527168

DeepFruits: A Fruit Detection System Using Deep Neural Networks.

PubMed

Sa, Inkyu; Ge, Zongyuan; Dayoub, Feras; Upcroft, Ben; Perez, Tristan; McCool, Chris

2016-08-03

This paper presents a novel approach to fruit detection using deep convolutional neural networks. The aim is to build an accurate, fast and reliable fruit detection system, which is a vital element of an autonomous agricultural robotic platform; it is a key element for fruit yield estimation and automated harvesting. Recent work in deep neural networks has led to the development of a state-of-the-art object detector termed Faster Region-based CNN (Faster R-CNN). We adapt this model, through transfer learning, for the task of fruit detection using imagery obtained from two modalities: colour (RGB) and Near-Infrared (NIR). Early and late fusion methods are explored for combining the multi-modal (RGB and NIR) information. This leads to a novel multi-modal Faster R-CNN model, which achieves state-of-the-art results compared to prior work with the F1 score, which takes into account both precision and recall performances improving from 0 . 807 to 0 . 838 for the detection of sweet pepper. In addition to improved accuracy, this approach is also much quicker to deploy for new fruits, as it requires bounding box annotation rather than pixel-level annotation (annotating bounding boxes is approximately an order of magnitude quicker to perform). The model is retrained to perform the detection of seven fruits, with the entire process taking four hours to annotate and train the new model per fruit.

Using deep neural networks to augment NIF post-shot analysis

NASA Astrophysics Data System (ADS)

Humbird, Kelli; Peterson, Luc; McClarren, Ryan; Field, John; Gaffney, Jim; Kruse, Michael; Nora, Ryan; Spears, Brian

2017-10-01

Post-shot analysis of National Ignition Facility (NIF) experiments is the process of determining which simulation inputs yield results consistent with experimental observations. This analysis is typically accomplished by running suites of manually adjusted simulations, or Monte Carlo sampling surrogate models that approximate the response surfaces of the physics code. These approaches are expensive and often find simulations that match only a small subset of observables simultaneously. We demonstrate an alternative method for performing post-shot analysis using inverse models, which map directly from experimental observables to simulation inputs with quantified uncertainties. The models are created using a novel machine learning algorithm which automates the construction and initialization of deep neural networks to optimize predictive accuracy. We show how these neural networks, trained on large databases of post-shot simulations, can rigorously quantify the agreement between simulation and experiment. This work performed under the auspices of the U.S. Department of Energy by Lawrence Livermore National Laboratory under Contract DE-AC52-07NA27344.

DeepID-Net: Deformable Deep Convolutional Neural Networks for Object Detection.

PubMed

Ouyang, Wanli; Zeng, Xingyu; Wang, Xiaogang; Qiu, Shi; Luo, Ping; Tian, Yonglong; Li, Hongsheng; Yang, Shuo; Wang, Zhe; Li, Hongyang; Loy, Chen Change; Wang, Kun; Yan, Junjie; Tang, Xiaoou

2016-07-07

In this paper, we propose deformable deep convolutional neural networks for generic object detection. This new deep learning object detection framework has innovations in multiple aspects. In the proposed new deep architecture, a new deformation constrained pooling (def-pooling) layer models the deformation of object parts with geometric constraint and penalty. A new pre-training strategy is proposed to learn feature representations more suitable for the object detection task and with good generalization capability. By changing the net structures, training strategies, adding and removing some key components in the detection pipeline, a set of models with large diversity are obtained, which significantly improves the effectiveness of model averaging. The proposed approach improves the mean averaged precision obtained by RCNN [16], which was the state-of-the-art, from 31% to 50.3% on the ILSVRC2014 detection test set. It also outperforms the winner of ILSVRC2014, GoogLeNet, by 6.1%. Detailed component-wise analysis is also provided through extensive experimental evaluation, which provides a global view for people to understand the deep learning object detection pipeline.

Combining morphometric features and convolutional networks fusion for glaucoma diagnosis

NASA Astrophysics Data System (ADS)

Perdomo, Oscar; Arevalo, John; González, Fabio A.

2017-11-01

Glaucoma is an eye condition that leads to loss of vision and blindness. Ophthalmoscopy exam evaluates the shape, color and proportion between the optic disc and physiologic cup, but the lack of agreement among experts is still the main diagnosis problem. The application of deep convolutional neural networks combined with automatic extraction of features such as: the cup-to-disc distance in the four quadrants, the perimeter, area, eccentricity, the major radio, the minor radio in optic disc and cup, in addition to all the ratios among the previous parameters may help with a better automatic grading of glaucoma. This paper presents a strategy to merge morphological features and deep convolutional neural networks as a novel methodology to support the glaucoma diagnosis in eye fundus images.

Radio frequency interference mitigation using deep convolutional neural networks

NASA Astrophysics Data System (ADS)

Akeret, J.; Chang, C.; Lucchi, A.; Refregier, A.

2017-01-01

We propose a novel approach for mitigating radio frequency interference (RFI) signals in radio data using the latest advances in deep learning. We employ a special type of Convolutional Neural Network, the U-Net, that enables the classification of clean signal and RFI signatures in 2D time-ordered data acquired from a radio telescope. We train and assess the performance of this network using the HIDE &SEEK radio data simulation and processing packages, as well as early Science Verification data acquired with the 7m single-dish telescope at the Bleien Observatory. We find that our U-Net implementation is showing competitive accuracy to classical RFI mitigation algorithms such as SEEK's SUMTHRESHOLD implementation. We publish our U-Net software package on GitHub under GPLv3 license.

Deep learning with coherent nanophotonic circuits

NASA Astrophysics Data System (ADS)

Shen, Yichen; Harris, Nicholas C.; Skirlo, Scott; Prabhu, Mihika; Baehr-Jones, Tom; Hochberg, Michael; Sun, Xin; Zhao, Shijie; Larochelle, Hugo; Englund, Dirk; Soljačić, Marin

2017-07-01

Artificial neural networks are computational network models inspired by signal processing in the brain. These models have dramatically improved performance for many machine-learning tasks, including speech and image recognition. However, today's computing hardware is inefficient at implementing neural networks, in large part because much of it was designed for von Neumann computing schemes. Significant effort has been made towards developing electronic architectures tuned to implement artificial neural networks that exhibit improved computational speed and accuracy. Here, we propose a new architecture for a fully optical neural network that, in principle, could offer an enhancement in computational speed and power efficiency over state-of-the-art electronics for conventional inference tasks. We experimentally demonstrate the essential part of the concept using a programmable nanophotonic processor featuring a cascaded array of 56 programmable Mach-Zehnder interferometers in a silicon photonic integrated circuit and show its utility for vowel recognition.

Prototype-Incorporated Emotional Neural Network.

PubMed

Oyedotun, Oyebade K; Khashman, Adnan

2017-08-15

Artificial neural networks (ANNs) aim to simulate the biological neural activities. Interestingly, many ''engineering'' prospects in ANN have relied on motivations from cognition and psychology studies. So far, two important learning theories that have been subject of active research are the prototype and adaptive learning theories. The learning rules employed for ANNs can be related to adaptive learning theory, where several examples of the different classes in a task are supplied to the network for adjusting internal parameters. Conversely, the prototype-learning theory uses prototypes (representative examples); usually, one prototype per class of the different classes contained in the task. These prototypes are supplied for systematic matching with new examples so that class association can be achieved. In this paper, we propose and implement a novel neural network algorithm based on modifying the emotional neural network (EmNN) model to unify the prototype- and adaptive-learning theories. We refer to our new model as ``prototype-incorporated EmNN''. Furthermore, we apply the proposed model to two real-life challenging tasks, namely, static hand-gesture recognition and face recognition, and compare the result to those obtained using the popular back-propagation neural network (BPNN), emotional BPNN (EmNN), deep networks, an exemplar classification model, and k-nearest neighbor.

Deblurring adaptive optics retinal images using deep convolutional neural networks.

PubMed

Fei, Xiao; Zhao, Junlei; Zhao, Haoxin; Yun, Dai; Zhang, Yudong

2017-12-01

The adaptive optics (AO) can be used to compensate for ocular aberrations to achieve near diffraction limited high-resolution retinal images. However, many factors such as the limited aberration measurement and correction accuracy with AO, intraocular scatter, imaging noise and so on will degrade the quality of retinal images. Image post processing is an indispensable and economical method to make up for the limitation of AO retinal imaging procedure. In this paper, we proposed a deep learning method to restore the degraded retinal images for the first time. The method directly learned an end-to-end mapping between the blurred and restored retinal images. The mapping was represented as a deep convolutional neural network that was trained to output high-quality images directly from blurry inputs without any preprocessing. This network was validated on synthetically generated retinal images as well as real AO retinal images. The assessment of the restored retinal images demonstrated that the image quality had been significantly improved.

Deblurring adaptive optics retinal images using deep convolutional neural networks

PubMed Central

Fei, Xiao; Zhao, Junlei; Zhao, Haoxin; Yun, Dai; Zhang, Yudong

2017-01-01

The adaptive optics (AO) can be used to compensate for ocular aberrations to achieve near diffraction limited high-resolution retinal images. However, many factors such as the limited aberration measurement and correction accuracy with AO, intraocular scatter, imaging noise and so on will degrade the quality of retinal images. Image post processing is an indispensable and economical method to make up for the limitation of AO retinal imaging procedure. In this paper, we proposed a deep learning method to restore the degraded retinal images for the first time. The method directly learned an end-to-end mapping between the blurred and restored retinal images. The mapping was represented as a deep convolutional neural network that was trained to output high-quality images directly from blurry inputs without any preprocessing. This network was validated on synthetically generated retinal images as well as real AO retinal images. The assessment of the restored retinal images demonstrated that the image quality had been significantly improved. PMID:29296496

Joint OSNR monitoring and modulation format identification in digital coherent receivers using deep neural networks.

PubMed

Khan, Faisal Nadeem; Zhong, Kangping; Zhou, Xian; Al-Arashi, Waled Hussein; Yu, Changyuan; Lu, Chao; Lau, Alan Pak Tao

2017-07-24

We experimentally demonstrate the use of deep neural networks (DNNs) in combination with signals' amplitude histograms (AHs) for simultaneous optical signal-to-noise ratio (OSNR) monitoring and modulation format identification (MFI) in digital coherent receivers. The proposed technique automatically extracts OSNR and modulation format dependent features of AHs, obtained after constant modulus algorithm (CMA) equalization, and exploits them for the joint estimation of these parameters. Experimental results for 112 Gbps polarization-multiplexed (PM) quadrature phase-shift keying (QPSK), 112 Gbps PM 16 quadrature amplitude modulation (16-QAM), and 240 Gbps PM 64-QAM signals demonstrate OSNR monitoring with mean estimation errors of 1.2 dB, 0.4 dB, and 1 dB, respectively. Similarly, the results for MFI show 100% identification accuracy for all three modulation formats. The proposed technique applies deep machine learning algorithms inside standard digital coherent receiver and does not require any additional hardware. Therefore, it is attractive for cost-effective multi-parameter estimation in next-generation elastic optical networks (EONs).

Phylogenetic convolutional neural networks in metagenomics.

PubMed

Fioravanti, Diego; Giarratano, Ylenia; Maggio, Valerio; Agostinelli, Claudio; Chierici, Marco; Jurman, Giuseppe; Furlanello, Cesare

2018-03-08

Convolutional Neural Networks can be effectively used only when data are endowed with an intrinsic concept of neighbourhood in the input space, as is the case of pixels in images. We introduce here Ph-CNN, a novel deep learning architecture for the classification of metagenomics data based on the Convolutional Neural Networks, with the patristic distance defined on the phylogenetic tree being used as the proximity measure. The patristic distance between variables is used together with a sparsified version of MultiDimensional Scaling to embed the phylogenetic tree in a Euclidean space. Ph-CNN is tested with a domain adaptation approach on synthetic data and on a metagenomics collection of gut microbiota of 38 healthy subjects and 222 Inflammatory Bowel Disease patients, divided in 6 subclasses. Classification performance is promising when compared to classical algorithms like Support Vector Machines and Random Forest and a baseline fully connected neural network, e.g. the Multi-Layer Perceptron. Ph-CNN represents a novel deep learning approach for the classification of metagenomics data. Operatively, the algorithm has been implemented as a custom Keras layer taking care of passing to the following convolutional layer not only the data but also the ranked list of neighbourhood of each sample, thus mimicking the case of image data, transparently to the user.

Rain/No-Rain Identification from Bispectral Satellite Information using Deep Neural Networks

NASA Astrophysics Data System (ADS)

Tao, Y.

2016-12-01

Satellite-based precipitation estimation products have the advantage of high resolution and global coverage. However, they still suffer from insufficient accuracy. To accurately estimate precipitation from satellite data, there are two most important aspects: sufficient precipitation information in the satellite information and proper methodologies to extract such information effectively. This study applies the state-of-the-art machine learning methodologies to bispectral satellite information for Rain/No-Rain detection. Specifically, we use deep neural networks to extract features from infrared and water vapor channels and connect it to precipitation identification. To evaluate the effectiveness of the methodology, we first applies it to the infrared data only (Model DL-IR only), the most commonly used inputs for satellite-based precipitation estimation. Then we incorporates water vapor data (Model DL-IR + WV) to further improve the prediction performance. Radar stage IV dataset is used as ground measurement for parameter calibration. The operational product, Precipitation Estimation from Remotely Sensed Information Using Artificial Neural Networks Cloud Classification System (PERSIANN-CCS), is used as a reference to compare the performance of both models in both winter and summer seasons.The experiments show significant improvement for both models in precipitation identification. The overall performance gains in the Critical Success Index (CSI) are 21.60% and 43.66% over the verification periods for Model DL-IR only and Model DL-IR+WV model compared to PERSIANN-CCS, respectively. Moreover, specific case studies show that the water vapor channel information and the deep neural networks effectively help recover a large number of missing precipitation pixels under warm clouds while reducing false alarms under cold clouds.

Framework for Detection and Localization of Extreme Climate Event with Pixel Recursive Super Resolution

NASA Astrophysics Data System (ADS)

Kim, S. K.; Lee, J.; Zhang, C.; Ames, S.; Williams, D. N.

2017-12-01

Deep learning techniques have been successfully applied to solve many problems in climate and geoscience using massive-scaled observed and modeled data. For extreme climate event detections, several models based on deep neural networks have been recently proposed and attend superior performance that overshadows all previous handcrafted expert based method. The issue arising, though, is that accurate localization of events requires high quality of climate data. In this work, we propose framework capable of detecting and localizing extreme climate events in very coarse climate data. Our framework is based on two models using deep neural networks, (1) Convolutional Neural Networks (CNNs) to detect and localize extreme climate events, and (2) Pixel recursive recursive super resolution model to reconstruct high resolution climate data from low resolution climate data. Based on our preliminary work, we have presented two CNNs in our framework for different purposes, detection and localization. Our results using CNNs for extreme climate events detection shows that simple neural nets can capture the pattern of extreme climate events with high accuracy from very coarse reanalysis data. However, localization accuracy is relatively low due to the coarse resolution. To resolve this issue, the pixel recursive super resolution model reconstructs the resolution of input of localization CNNs. We present a best networks using pixel recursive super resolution model that synthesizes details of tropical cyclone in ground truth data while enhancing their resolution. Therefore, this approach not only dramat- ically reduces the human effort, but also suggests possibility to reduce computing cost required for downscaling process to increase resolution of data.

Planning chemical syntheses with deep neural networks and symbolic AI

NASA Astrophysics Data System (ADS)

Segler, Marwin H. S.; Preuss, Mike; Waller, Mark P.

2018-03-01

To plan the syntheses of small organic molecules, chemists use retrosynthesis, a problem-solving technique in which target molecules are recursively transformed into increasingly simpler precursors. Computer-aided retrosynthesis would be a valuable tool but at present it is slow and provides results of unsatisfactory quality. Here we use Monte Carlo tree search and symbolic artificial intelligence (AI) to discover retrosynthetic routes. We combined Monte Carlo tree search with an expansion policy network that guides the search, and a filter network to pre-select the most promising retrosynthetic steps. These deep neural networks were trained on essentially all reactions ever published in organic chemistry. Our system solves for almost twice as many molecules, thirty times faster than the traditional computer-aided search method, which is based on extracted rules and hand-designed heuristics. In a double-blind AB test, chemists on average considered our computer-generated routes to be equivalent to reported literature routes.

Improving deep convolutional neural networks with mixed maxout units.

PubMed

Zhao, Hui-Zhen; Liu, Fu-Xian; Li, Long-Yue

2017-01-01

Motivated by insights from the maxout-units-based deep Convolutional Neural Network (CNN) that "non-maximal features are unable to deliver" and "feature mapping subspace pooling is insufficient," we present a novel mixed variant of the recently introduced maxout unit called a mixout unit. Specifically, we do so by calculating the exponential probabilities of feature mappings gained by applying different convolutional transformations over the same input and then calculating the expected values according to their exponential probabilities. Moreover, we introduce the Bernoulli distribution to balance the maximum values with the expected values of the feature mappings subspace. Finally, we design a simple model to verify the pooling ability of mixout units and a Mixout-units-based Network-in-Network (NiN) model to analyze the feature learning ability of the mixout models. We argue that our proposed units improve the pooling ability and that mixout models can achieve better feature learning and classification performance.

Depth Reconstruction from Single Images Using a Convolutional Neural Network and a Condition Random Field Model.

PubMed

Liu, Dan; Liu, Xuejun; Wu, Yiguang

2018-04-24

This paper presents an effective approach for depth reconstruction from a single image through the incorporation of semantic information and local details from the image. A unified framework for depth acquisition is constructed by joining a deep Convolutional Neural Network (CNN) and a continuous pairwise Conditional Random Field (CRF) model. Semantic information and relative depth trends of local regions inside the image are integrated into the framework. A deep CNN network is firstly used to automatically learn a hierarchical feature representation of the image. To get more local details in the image, the relative depth trends of local regions are incorporated into the network. Combined with semantic information of the image, a continuous pairwise CRF is then established and is used as the loss function of the unified model. Experiments on real scenes demonstrate that the proposed approach is effective and that the approach obtains satisfactory results.

Deep learning in breast cancer risk assessment: evaluation of convolutional neural networks on a clinical dataset of full-field digital mammograms.

PubMed

Li, Hui; Giger, Maryellen L; Huynh, Benjamin Q; Antropova, Natalia O

2017-10-01

To evaluate deep learning in the assessment of breast cancer risk in which convolutional neural networks (CNNs) with transfer learning are used to extract parenchymal characteristics directly from full-field digital mammographic (FFDM) images instead of using computerized radiographic texture analysis (RTA), 456 clinical FFDM cases were included: a "high-risk" BRCA1/2 gene-mutation carriers dataset (53 cases), a "high-risk" unilateral cancer patients dataset (75 cases), and a "low-risk dataset" (328 cases). Deep learning was compared to the use of features from RTA, as well as to a combination of both in the task of distinguishing between high- and low-risk subjects. Similar classification performances were obtained using CNN [area under the curve [Formula: see text]; standard error [Formula: see text

Completing sparse and disconnected protein-protein network by deep learning.

PubMed

Huang, Lei; Liao, Li; Wu, Cathy H

2018-03-22

Protein-protein interaction (PPI) prediction remains a central task in systems biology to achieve a better and holistic understanding of cellular and intracellular processes. Recently, an increasing number of computational methods have shifted from pair-wise prediction to network level prediction. Many of the existing network level methods predict PPIs under the assumption that the training network should be connected. However, this assumption greatly affects the prediction power and limits the application area because the current golden standard PPI networks are usually very sparse and disconnected. Therefore, how to effectively predict PPIs based on a training network that is sparse and disconnected remains a challenge. In this work, we developed a novel PPI prediction method based on deep learning neural network and regularized Laplacian kernel. We use a neural network with an autoencoder-like architecture to implicitly simulate the evolutionary processes of a PPI network. Neurons of the output layer correspond to proteins and are labeled with values (1 for interaction and 0 for otherwise) from the adjacency matrix of a sparse disconnected training PPI network. Unlike autoencoder, neurons at the input layer are given all zero input, reflecting an assumption of no a priori knowledge about PPIs, and hidden layers of smaller sizes mimic ancient interactome at different times during evolution. After the training step, an evolved PPI network whose rows are outputs of the neural network can be obtained. We then predict PPIs by applying the regularized Laplacian kernel to the transition matrix that is built upon the evolved PPI network. The results from cross-validation experiments show that the PPI prediction accuracies for yeast data and human data measured as AUC are increased by up to 8.4 and 14.9% respectively, as compared to the baseline. Moreover, the evolved PPI network can also help us leverage complementary information from the disconnected training network and multiple heterogeneous data sources. Tested by the yeast data with six heterogeneous feature kernels, the results show our method can further improve the prediction performance by up to 2%, which is very close to an upper bound that is obtained by an Approximate Bayesian Computation based sampling method. The proposed evolution deep neural network, coupled with regularized Laplacian kernel, is an effective tool in completing sparse and disconnected PPI networks and in facilitating integration of heterogeneous data sources.

Deep Neural Network Detects Quantum Phase Transition

NASA Astrophysics Data System (ADS)

Arai, Shunta; Ohzeki, Masayuki; Tanaka, Kazuyuki

2018-03-01

We detect the quantum phase transition of a quantum many-body system by mapping the observed results of the quantum state onto a neural network. In the present study, we utilized the simplest case of a quantum many-body system, namely a one-dimensional chain of Ising spins with the transverse Ising model. We prepared several spin configurations, which were obtained using repeated observations of the model for a particular strength of the transverse field, as input data for the neural network. Although the proposed method can be employed using experimental observations of quantum many-body systems, we tested our technique with spin configurations generated by a quantum Monte Carlo simulation without initial relaxation. The neural network successfully identified the strength of transverse field only from the spin configurations, leading to consistent estimations of the critical point of our model Γc = J.

Deep Convolutional Neural Networks for Endotracheal Tube Position and X-ray Image Classification: Challenges and Opportunities.

PubMed

Lakhani, Paras

2017-08-01

The goal of this study is to evaluate the efficacy of deep convolutional neural networks (DCNNs) in differentiating subtle, intermediate, and more obvious image differences in radiography. Three different datasets were created, which included presence/absence of the endotracheal (ET) tube (n = 300), low/normal position of the ET tube (n = 300), and chest/abdominal radiographs (n = 120). The datasets were split into training, validation, and test. Both untrained and pre-trained deep neural networks were employed, including AlexNet and GoogLeNet classifiers, using the Caffe framework. Data augmentation was performed for the presence/absence and low/normal ET tube datasets. Receiver operating characteristic (ROC), area under the curves (AUC), and 95% confidence intervals were calculated. Statistical differences of the AUCs were determined using a non-parametric approach. The pre-trained AlexNet and GoogLeNet classifiers had perfect accuracy (AUC 1.00) in differentiating chest vs. abdominal radiographs, using only 45 training cases. For more difficult datasets, including the presence/absence and low/normal position endotracheal tubes, more training cases, pre-trained networks, and data-augmentation approaches were helpful to increase accuracy. The best-performing network for classifying presence vs. absence of an ET tube was still very accurate with an AUC of 0.99. However, for the most difficult dataset, such as low vs. normal position of the endotracheal tube, DCNNs did not perform as well, but achieved a reasonable AUC of 0.81.

Deep convolutional neural networks for classifying GPR B-scans

NASA Astrophysics Data System (ADS)

Besaw, Lance E.; Stimac, Philip J.

2015-05-01

Symmetric and asymmetric buried explosive hazards (BEHs) present real, persistent, deadly threats on the modern battlefield. Current approaches to mitigate these threats rely on highly trained operatives to reliably detect BEHs with reasonable false alarm rates using handheld Ground Penetrating Radar (GPR) and metal detectors. As computers become smaller, faster and more efficient, there exists greater potential for automated threat detection based on state-of-the-art machine learning approaches, reducing the burden on the field operatives. Recent advancements in machine learning, specifically deep learning artificial neural networks, have led to significantly improved performance in pattern recognition tasks, such as object classification in digital images. Deep convolutional neural networks (CNNs) are used in this work to extract meaningful signatures from 2-dimensional (2-D) GPR B-scans and classify threats. The CNNs skip the traditional "feature engineering" step often associated with machine learning, and instead learn the feature representations directly from the 2-D data. A multi-antennae, handheld GPR with centimeter-accurate positioning data was used to collect shallow subsurface data over prepared lanes containing a wide range of BEHs. Several heuristics were used to prevent over-training, including cross validation, network weight regularization, and "dropout." Our results show that CNNs can extract meaningful features and accurately classify complex signatures contained in GPR B-scans, complementing existing GPR feature extraction and classification techniques.

Deep Neural Networks Reveal a Gradient in the Complexity of Neural Representations across the Ventral Stream.

PubMed

Güçlü, Umut; van Gerven, Marcel A J

2015-07-08

Converging evidence suggests that the primate ventral visual pathway encodes increasingly complex stimulus features in downstream areas. We quantitatively show that there indeed exists an explicit gradient for feature complexity in the ventral pathway of the human brain. This was achieved by mapping thousands of stimulus features of increasing complexity across the cortical sheet using a deep neural network. Our approach also revealed a fine-grained functional specialization of downstream areas of the ventral stream. Furthermore, it allowed decoding of representations from human brain activity at an unsurpassed degree of accuracy, confirming the quality of the developed approach. Stimulus features that successfully explained neural responses indicate that population receptive fields were explicitly tuned for object categorization. This provides strong support for the hypothesis that object categorization is a guiding principle in the functional organization of the primate ventral stream. Copyright © 2015 the authors 0270-6474/15/3510005-10$15.00/0.

Fabric defect detection based on visual saliency using deep feature and low-rank recovery

NASA Astrophysics Data System (ADS)

Liu, Zhoufeng; Wang, Baorui; Li, Chunlei; Li, Bicao; Dong, Yan

2018-04-01

Fabric defect detection plays an important role in improving the quality of fabric product. In this paper, a novel fabric defect detection method based on visual saliency using deep feature and low-rank recovery was proposed. First, unsupervised training is carried out by the initial network parameters based on MNIST large datasets. The supervised fine-tuning of fabric image library based on Convolutional Neural Networks (CNNs) is implemented, and then more accurate deep neural network model is generated. Second, the fabric images are uniformly divided into the image block with the same size, then we extract their multi-layer deep features using the trained deep network. Thereafter, all the extracted features are concentrated into a feature matrix. Third, low-rank matrix recovery is adopted to divide the feature matrix into the low-rank matrix which indicates the background and the sparse matrix which indicates the salient defect. In the end, the iterative optimal threshold segmentation algorithm is utilized to segment the saliency maps generated by the sparse matrix to locate the fabric defect area. Experimental results demonstrate that the feature extracted by CNN is more suitable for characterizing the fabric texture than the traditional LBP, HOG and other hand-crafted features extraction method, and the proposed method can accurately detect the defect regions of various fabric defects, even for the image with complex texture.

Deep convolutional neural networks for classifying head and neck cancer using hyperspectral imaging

NASA Astrophysics Data System (ADS)

Halicek, Martin; Lu, Guolan; Little, James V.; Wang, Xu; Patel, Mihir; Griffith, Christopher C.; El-Deiry, Mark W.; Chen, Amy Y.; Fei, Baowei

2017-06-01

Surgical cancer resection requires an accurate and timely diagnosis of the cancer margins in order to achieve successful patient remission. Hyperspectral imaging (HSI) has emerged as a useful, noncontact technique for acquiring spectral and optical properties of tissue. A convolutional neural network (CNN) classifier is developed to classify excised, squamous-cell carcinoma, thyroid cancer, and normal head and neck tissue samples using HSI. The CNN classification was validated by the manual annotation of a pathologist specialized in head and neck cancer. The preliminary results of 50 patients indicate the potential of HSI and deep learning for automatic tissue-labeling of surgical specimens of head and neck patients.

Deep neural network and noise classification-based speech enhancement

NASA Astrophysics Data System (ADS)

Shi, Wenhua; Zhang, Xiongwei; Zou, Xia; Han, Wei

2017-07-01

In this paper, a speech enhancement method using noise classification and Deep Neural Network (DNN) was proposed. Gaussian mixture model (GMM) was employed to determine the noise type in speech-absent frames. DNN was used to model the relationship between noisy observation and clean speech. Once the noise type was determined, the corresponding DNN model was applied to enhance the noisy speech. GMM was trained with mel-frequency cepstrum coefficients (MFCC) and the parameters were estimated with an iterative expectation-maximization (EM) algorithm. Noise type was updated by spectrum entropy-based voice activity detection (VAD). Experimental results demonstrate that the proposed method could achieve better objective speech quality and smaller distortion under stationary and non-stationary conditions.

Deep Unfolding for Topic Models.

PubMed

Chien, Jen-Tzung; Lee, Chao-Hsi

2018-02-01

Deep unfolding provides an approach to integrate the probabilistic generative models and the deterministic neural networks. Such an approach is benefited by deep representation, easy interpretation, flexible learning and stochastic modeling. This study develops the unsupervised and supervised learning of deep unfolded topic models for document representation and classification. Conventionally, the unsupervised and supervised topic models are inferred via the variational inference algorithm where the model parameters are estimated by maximizing the lower bound of logarithm of marginal likelihood using input documents without and with class labels, respectively. The representation capability or classification accuracy is constrained by the variational lower bound and the tied model parameters across inference procedure. This paper aims to relax these constraints by directly maximizing the end performance criterion and continuously untying the parameters in learning process via deep unfolding inference (DUI). The inference procedure is treated as the layer-wise learning in a deep neural network. The end performance is iteratively improved by using the estimated topic parameters according to the exponentiated updates. Deep learning of topic models is therefore implemented through a back-propagation procedure. Experimental results show the merits of DUI with increasing number of layers compared with variational inference in unsupervised as well as supervised topic models.

Predicting healthcare trajectories from medical records: A deep learning approach.

PubMed

Pham, Trang; Tran, Truyen; Phung, Dinh; Venkatesh, Svetha

2017-05-01

Personalized predictive medicine necessitates the modeling of patient illness and care processes, which inherently have long-term temporal dependencies. Healthcare observations, stored in electronic medical records are episodic and irregular in time. We introduce DeepCare, an end-to-end deep dynamic neural network that reads medical records, stores previous illness history, infers current illness states and predicts future medical outcomes. At the data level, DeepCare represents care episodes as vectors and models patient health state trajectories by the memory of historical records. Built on Long Short-Term Memory (LSTM), DeepCare introduces methods to handle irregularly timed events by moderating the forgetting and consolidation of memory. DeepCare also explicitly models medical interventions that change the course of illness and shape future medical risk. Moving up to the health state level, historical and present health states are then aggregated through multiscale temporal pooling, before passing through a neural network that estimates future outcomes. We demonstrate the efficacy of DeepCare for disease progression modeling, intervention recommendation, and future risk prediction. On two important cohorts with heavy social and economic burden - diabetes and mental health - the results show improved prediction accuracy. Copyright © 2017 Elsevier Inc. All rights reserved.

Visual pathways from the perspective of cost functions and multi-task deep neural networks.

PubMed

Scholte, H Steven; Losch, Max M; Ramakrishnan, Kandan; de Haan, Edward H F; Bohte, Sander M

2018-01-01

Vision research has been shaped by the seminal insight that we can understand the higher-tier visual cortex from the perspective of multiple functional pathways with different goals. In this paper, we try to give a computational account of the functional organization of this system by reasoning from the perspective of multi-task deep neural networks. Machine learning has shown that tasks become easier to solve when they are decomposed into subtasks with their own cost function. We hypothesize that the visual system optimizes multiple cost functions of unrelated tasks and this causes the emergence of a ventral pathway dedicated to vision for perception, and a dorsal pathway dedicated to vision for action. To evaluate the functional organization in multi-task deep neural networks, we propose a method that measures the contribution of a unit towards each task, applying it to two networks that have been trained on either two related or two unrelated tasks, using an identical stimulus set. Results show that the network trained on the unrelated tasks shows a decreasing degree of feature representation sharing towards higher-tier layers while the network trained on related tasks uniformly shows high degree of sharing. We conjecture that the method we propose can be used to analyze the anatomical and functional organization of the visual system and beyond. We predict that the degree to which tasks are related is a good descriptor of the degree to which they share downstream cortical-units. Copyright © 2017 Elsevier Ltd. All rights reserved.

Fine-grained leukocyte classification with deep residual learning for microscopic images.

PubMed

Qin, Feiwei; Gao, Nannan; Peng, Yong; Wu, Zizhao; Shen, Shuying; Grudtsin, Artur

2018-08-01

Leukocyte classification and cytometry have wide applications in medical domain, previous researches usually exploit machine learning techniques to classify leukocytes automatically. However, constrained by the past development of machine learning techniques, for example, extracting distinctive features from raw microscopic images are difficult, the widely used SVM classifier only has relative few parameters to tune, these methods cannot efficiently handle fine-grained classification cases when the white blood cells have up to 40 categories. Based on deep learning theory, a systematic study is conducted on finer leukocyte classification in this paper. A deep residual neural network based leukocyte classifier is constructed at first, which can imitate the domain expert's cell recognition process, and extract salient features robustly and automatically. Then the deep neural network classifier's topology is adjusted according to the prior knowledge of white blood cell test. After that the microscopic image dataset with almost one hundred thousand labeled leukocytes belonging to 40 categories is built, and combined training strategies are adopted to make the designed classifier has good generalization ability. The proposed deep residual neural network based classifier was tested on microscopic image dataset with 40 leukocyte categories. It achieves top-1 accuracy of 77.80%, top-5 accuracy of 98.75% during the training procedure. The average accuracy on the test set is nearly 76.84%. This paper presents a fine-grained leukocyte classification method for microscopic images, based on deep residual learning theory and medical domain knowledge. Experimental results validate the feasibility and effectiveness of our approach. Extended experiments support that the fine-grained leukocyte classifier could be used in real medical applications, assist doctors in diagnosing diseases, reduce human power significantly. Copyright © 2018 Elsevier B.V. All rights reserved.

Automatically identifying, counting, and describing wild animals in camera-trap images with deep learning.

PubMed

Norouzzadeh, Mohammad Sadegh; Nguyen, Anh; Kosmala, Margaret; Swanson, Alexandra; Palmer, Meredith S; Packer, Craig; Clune, Jeff

2018-06-19

Having accurate, detailed, and up-to-date information about the location and behavior of animals in the wild would improve our ability to study and conserve ecosystems. We investigate the ability to automatically, accurately, and inexpensively collect such data, which could help catalyze the transformation of many fields of ecology, wildlife biology, zoology, conservation biology, and animal behavior into "big data" sciences. Motion-sensor "camera traps" enable collecting wildlife pictures inexpensively, unobtrusively, and frequently. However, extracting information from these pictures remains an expensive, time-consuming, manual task. We demonstrate that such information can be automatically extracted by deep learning, a cutting-edge type of artificial intelligence. We train deep convolutional neural networks to identify, count, and describe the behaviors of 48 species in the 3.2 million-image Snapshot Serengeti dataset. Our deep neural networks automatically identify animals with >93.8% accuracy, and we expect that number to improve rapidly in years to come. More importantly, if our system classifies only images it is confident about, our system can automate animal identification for 99.3% of the data while still performing at the same 96.6% accuracy as that of crowdsourced teams of human volunteers, saving >8.4 y (i.e., >17,000 h at 40 h/wk) of human labeling effort on this 3.2 million-image dataset. Those efficiency gains highlight the importance of using deep neural networks to automate data extraction from camera-trap images, reducing a roadblock for this widely used technology. Our results suggest that deep learning could enable the inexpensive, unobtrusive, high-volume, and even real-time collection of a wealth of information about vast numbers of animals in the wild. Copyright © 2018 the Author(s). Published by PNAS.

Face recognition via Gabor and convolutional neural network

NASA Astrophysics Data System (ADS)

Lu, Tongwei; Wu, Menglu; Lu, Tao

2018-04-01

In recent years, the powerful feature learning and classification ability of convolutional neural network have attracted widely attention. Compared with the deep learning, the traditional machine learning algorithm has a good explanatory which deep learning does not have. Thus, In this paper, we propose a method to extract the feature of the traditional algorithm as the input of convolution neural network. In order to reduce the complexity of the network, the kernel function of Gabor wavelet is used to extract the feature from different position, frequency and direction of target image. It is sensitive to edge of image which can provide good direction and scale selection. The extraction of the image from eight directions on a scale are as the input of network that we proposed. The network have the advantage of weight sharing and local connection and texture feature of the input image can reduce the influence of facial expression, gesture and illumination. At the same time, we introduced a layer which combined the results of the pooling and convolution can extract deeper features. The training network used the open source caffe framework which is beneficial to feature extraction. The experiment results of the proposed method proved that the network structure effectively overcame the barrier of illumination and had a good robustness as well as more accurate and rapid than the traditional algorithm.

Real-time object-to-features vectorisation via Siamese neural networks

NASA Astrophysics Data System (ADS)

Fedorenko, Fedor; Usilin, Sergey

2017-03-01

Object-to-features vectorisation is a hard problem to solve for objects that can be hard to distinguish. Siamese and Triplet neural networks are one of the more recent tools used for such task. However, most networks used are very deep networks that prove to be hard to compute in the Internet of Things setting. In this paper, a computationally efficient neural network is proposed for real-time object-to-features vectorisation into a Euclidean metric space. We use L2 distance to reflect feature vector similarity during both training and testing. In this way, feature vectors we develop can be easily classified using K-Nearest Neighbours classifier. Such approach can be used to train networks to vectorise such "problematic" objects like images of human faces, keypoint image patches, like keypoints on Arctic maps and surrounding marine areas.

Hierarchical Recurrent Neural Hashing for Image Retrieval With Hierarchical Convolutional Features.

PubMed

Lu, Xiaoqiang; Chen, Yaxiong; Li, Xuelong

Hashing has been an important and effective technology in image retrieval due to its computational efficiency and fast search speed. The traditional hashing methods usually learn hash functions to obtain binary codes by exploiting hand-crafted features, which cannot optimally represent the information of the sample. Recently, deep learning methods can achieve better performance, since deep learning architectures can learn more effective image representation features. However, these methods only use semantic features to generate hash codes by shallow projection but ignore texture details. In this paper, we proposed a novel hashing method, namely hierarchical recurrent neural hashing (HRNH), to exploit hierarchical recurrent neural network to generate effective hash codes. There are three contributions of this paper. First, a deep hashing method is proposed to extensively exploit both spatial details and semantic information, in which, we leverage hierarchical convolutional features to construct image pyramid representation. Second, our proposed deep network can exploit directly convolutional feature maps as input to preserve the spatial structure of convolutional feature maps. Finally, we propose a new loss function that considers the quantization error of binarizing the continuous embeddings into the discrete binary codes, and simultaneously maintains the semantic similarity and balanceable property of hash codes. Experimental results on four widely used data sets demonstrate that the proposed HRNH can achieve superior performance over other state-of-the-art hashing methods.Hashing has been an important and effective technology in image retrieval due to its computational efficiency and fast search speed. The traditional hashing methods usually learn hash functions to obtain binary codes by exploiting hand-crafted features, which cannot optimally represent the information of the sample. Recently, deep learning methods can achieve better performance, since deep learning architectures can learn more effective image representation features. However, these methods only use semantic features to generate hash codes by shallow projection but ignore texture details. In this paper, we proposed a novel hashing method, namely hierarchical recurrent neural hashing (HRNH), to exploit hierarchical recurrent neural network to generate effective hash codes. There are three contributions of this paper. First, a deep hashing method is proposed to extensively exploit both spatial details and semantic information, in which, we leverage hierarchical convolutional features to construct image pyramid representation. Second, our proposed deep network can exploit directly convolutional feature maps as input to preserve the spatial structure of convolutional feature maps. Finally, we propose a new loss function that considers the quantization error of binarizing the continuous embeddings into the discrete binary codes, and simultaneously maintains the semantic similarity and balanceable property of hash codes. Experimental results on four widely used data sets demonstrate that the proposed HRNH can achieve superior performance over other state-of-the-art hashing methods.

Deep learning methods for protein torsion angle prediction.

PubMed

Li, Haiou; Hou, Jie; Adhikari, Badri; Lyu, Qiang; Cheng, Jianlin

2017-09-18

Deep learning is one of the most powerful machine learning methods that has achieved the state-of-the-art performance in many domains. Since deep learning was introduced to the field of bioinformatics in 2012, it has achieved success in a number of areas such as protein residue-residue contact prediction, secondary structure prediction, and fold recognition. In this work, we developed deep learning methods to improve the prediction of torsion (dihedral) angles of proteins. We design four different deep learning architectures to predict protein torsion angles. The architectures including deep neural network (DNN) and deep restricted Boltzmann machine (DRBN), deep recurrent neural network (DRNN) and deep recurrent restricted Boltzmann machine (DReRBM) since the protein torsion angle prediction is a sequence related problem. In addition to existing protein features, two new features (predicted residue contact number and the error distribution of torsion angles extracted from sequence fragments) are used as input to each of the four deep learning architectures to predict phi and psi angles of protein backbone. The mean absolute error (MAE) of phi and psi angles predicted by DRNN, DReRBM, DRBM and DNN is about 20-21° and 29-30° on an independent dataset. The MAE of phi angle is comparable to the existing methods, but the MAE of psi angle is 29°, 2° lower than the existing methods. On the latest CASP12 targets, our methods also achieved the performance better than or comparable to a state-of-the art method. Our experiment demonstrates that deep learning is a valuable method for predicting protein torsion angles. The deep recurrent network architecture performs slightly better than deep feed-forward architecture, and the predicted residue contact number and the error distribution of torsion angles extracted from sequence fragments are useful features for improving prediction accuracy.

Stimulated Deep Neural Network for Speech Recognition

DTIC Science & Technology

2016-09-08

making network regularization and robust adaptation challenging. Stimulated training has recently been proposed to address this problem by encouraging...potential to improve regularization and adaptation. This paper investigates stimulated training of DNNs for both of these options. These schemes take

Residual Deep Convolutional Neural Network Predicts MGMT Methylation Status.

PubMed

Korfiatis, Panagiotis; Kline, Timothy L; Lachance, Daniel H; Parney, Ian F; Buckner, Jan C; Erickson, Bradley J

2017-10-01

Predicting methylation of the O6-methylguanine methyltransferase (MGMT) gene status utilizing MRI imaging is of high importance since it is a predictor of response and prognosis in brain tumors. In this study, we compare three different residual deep neural network (ResNet) architectures to evaluate their ability in predicting MGMT methylation status without the need for a distinct tumor segmentation step. We found that the ResNet50 (50 layers) architecture was the best performing model, achieving an accuracy of 94.90% (+/- 3.92%) for the test set (classification of a slice as no tumor, methylated MGMT, or non-methylated). ResNet34 (34 layers) achieved 80.72% (+/- 13.61%) while ResNet18 (18 layers) accuracy was 76.75% (+/- 20.67%). ResNet50 performance was statistically significantly better than both ResNet18 and ResNet34 architectures (p < 0.001). We report a method that alleviates the need of extensive preprocessing and acts as a proof of concept that deep neural architectures can be used to predict molecular biomarkers from routine medical images.

Statistical downscaling of precipitation using long short-term memory recurrent neural networks

NASA Astrophysics Data System (ADS)

Misra, Saptarshi; Sarkar, Sudeshna; Mitra, Pabitra

2017-11-01

Hydrological impacts of global climate change on regional scale are generally assessed by downscaling large-scale climatic variables, simulated by General Circulation Models (GCMs), to regional, small-scale hydrometeorological variables like precipitation, temperature, etc. In this study, we propose a new statistical downscaling model based on Recurrent Neural Network with Long Short-Term Memory which captures the spatio-temporal dependencies in local rainfall. The previous studies have used several other methods such as linear regression, quantile regression, kernel regression, beta regression, and artificial neural networks. Deep neural networks and recurrent neural networks have been shown to be highly promising in modeling complex and highly non-linear relationships between input and output variables in different domains and hence we investigated their performance in the task of statistical downscaling. We have tested this model on two datasets—one on precipitation in Mahanadi basin in India and the second on precipitation in Campbell River basin in Canada. Our autoencoder coupled long short-term memory recurrent neural network model performs the best compared to other existing methods on both the datasets with respect to temporal cross-correlation, mean squared error, and capturing the extremes.

Nonvolatile Memory Materials for Neuromorphic Intelligent Machines.

PubMed

Jeong, Doo Seok; Hwang, Cheol Seong

2018-04-18

Recent progress in deep learning extends the capability of artificial intelligence to various practical tasks, making the deep neural network (DNN) an extremely versatile hypothesis. While such DNN is virtually built on contemporary data centers of the von Neumann architecture, physical (in part) DNN of non-von Neumann architecture, also known as neuromorphic computing, can remarkably improve learning and inference efficiency. Particularly, resistance-based nonvolatile random access memory (NVRAM) highlights its handy and efficient application to the multiply-accumulate (MAC) operation in an analog manner. Here, an overview is given of the available types of resistance-based NVRAMs and their technological maturity from the material- and device-points of view. Examples within the strategy are subsequently addressed in comparison with their benchmarks (virtual DNN in deep learning). A spiking neural network (SNN) is another type of neural network that is more biologically plausible than the DNN. The successful incorporation of resistance-based NVRAM in SNN-based neuromorphic computing offers an efficient solution to the MAC operation and spike timing-based learning in nature. This strategy is exemplified from a material perspective. Intelligent machines are categorized according to their architecture and learning type. Also, the functionality and usefulness of NVRAM-based neuromorphic computing are addressed. © 2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim.

Automatic detection of kidney in 3D pediatric ultrasound images using deep neural networks

NASA Astrophysics Data System (ADS)

Tabrizi, Pooneh R.; Mansoor, Awais; Biggs, Elijah; Jago, James; Linguraru, Marius George

2018-02-01

Ultrasound (US) imaging is the routine and safe diagnostic modality for detecting pediatric urology problems, such as hydronephrosis in the kidney. Hydronephrosis is the swelling of one or both kidneys because of the build-up of urine. Early detection of hydronephrosis can lead to a substantial improvement in kidney health outcomes. Generally, US imaging is a challenging modality for the evaluation of pediatric kidneys with different shape, size, and texture characteristics. The aim of this study is to present an automatic detection method to help kidney analysis in pediatric 3DUS images. The method localizes the kidney based on its minimum volume oriented bounding box) using deep neural networks. Separate deep neural networks are trained to estimate the kidney position, orientation, and scale, making the method computationally efficient by avoiding full parameter training. The performance of the method was evaluated using a dataset of 45 kidneys (18 normal and 27 diseased kidneys diagnosed with hydronephrosis) through the leave-one-out cross validation method. Quantitative results show the proposed detection method could extract the kidney position, orientation, and scale ratio with root mean square values of 1.3 +/- 0.9 mm, 6.34 +/- 4.32 degrees, and 1.73 +/- 0.04, respectively. This method could be helpful in automating kidney segmentation for routine clinical evaluation.

Automated embolic signal detection using Deep Convolutional Neural Network.

PubMed

Sombune, Praotasna; Phienphanich, Phongphan; Phuechpanpaisal, Sutanya; Muengtaweepongsa, Sombat; Ruamthanthong, Anuchit; Tantibundhit, Charturong

2017-07-01

This work investigated the potential of Deep Neural Network in detection of cerebral embolic signal (ES) from transcranial Doppler ultrasound (TCD). The resulting system is aimed to couple with TCD devices in diagnosing a risk of stroke in real-time with high accuracy. The Adaptive Gain Control (AGC) approach developed in our previous study is employed to capture suspected ESs in real-time. By using spectrograms of the same TCD signal dataset as that of our previous work as inputs and the same experimental setup, Deep Convolutional Neural Network (CNN), which can learn features while training, was investigated for its ability to bypass the traditional handcrafted feature extraction and selection process. Extracted feature vectors from the suspected ESs are later determined whether they are of an ES, artifact (AF) or normal (NR) interval. The effectiveness of the developed system was evaluated over 19 subjects going under procedures generating emboli. The CNN-based system could achieve in average of 83.0% sensitivity, 80.1% specificity, and 81.4% accuracy, with considerably much less time consumption in development. The certainly growing set of training samples and computational resources will contribute to high performance. Besides having potential use in various clinical ES monitoring settings, continuation of this promising study will benefit developments of wearable applications by leveraging learnable features to serve demographic differentials.

Detection and Prediction of Hail Storms in Satellite Imagery using Deep Learning

NASA Astrophysics Data System (ADS)

Pullman, M.; Gurung, I.; Ramachandran, R.; Maskey, M.

2017-12-01

Natural hazards, such as damaging hail storms, dramatically disrupt both industry and agriculture, having significant socio-economic impacts in the United States. In 2016, hail was responsible for 3.5 billion and 23 million dollars in damage to property and crops, respectively, making it the second costliest 2016 weather phenomenon in the United States. The destructive nature and high cost of hail storms has driven research into the development of more accurate hail-prediction algorithms in an effort to mitigate societal impacts. Recently, weather forecasting efforts have turned to deep learning neural networks because neural networks can more effectively model complex, nonlinear, dynamical phenomenon that exist in large datasets through multiple stages of transformation and representation. In an effort to improve hail-prediction techniques, we propose a deep learning technique that leverages satellite imagery to detect and predict the occurrence of hail storms. The technique is applied to satellite imagery from 2006 to 2016 for the contiguous United States and incorporates hail reports obtained from the National Center for Environmental Information Storm Events Database for training and validation purposes. In this presentation, we describe a novel approach to predicting hail via a neural network model that creates a large labeled dataset of hail storms, the accuracy and results of the model, and its applications for improving hail forecasting.

ANI-1: an extensible neural network potential with DFT accuracy at force field computational cost† †Electronic supplementary information (ESI) available. See DOI: 10.1039/c6sc05720a Click here for additional data file.

PubMed Central

Smith, J. S.

2017-01-01

Deep learning is revolutionizing many areas of science and technology, especially image, text, and speech recognition. In this paper, we demonstrate how a deep neural network (NN) trained on quantum mechanical (QM) DFT calculations can learn an accurate and transferable potential for organic molecules. We introduce ANAKIN-ME (Accurate NeurAl networK engINe for Molecular Energies) or ANI for short. ANI is a new method designed with the intent of developing transferable neural network potentials that utilize a highly-modified version of the Behler and Parrinello symmetry functions to build single-atom atomic environment vectors (AEV) as a molecular representation. AEVs provide the ability to train neural networks to data that spans both configurational and conformational space, a feat not previously accomplished on this scale. We utilized ANI to build a potential called ANI-1, which was trained on a subset of the GDB databases with up to 8 heavy atoms in order to predict total energies for organic molecules containing four atom types: H, C, N, and O. To obtain an accelerated but physically relevant sampling of molecular potential surfaces, we also proposed a Normal Mode Sampling (NMS) method for generating molecular conformations. Through a series of case studies, we show that ANI-1 is chemically accurate compared to reference DFT calculations on much larger molecular systems (up to 54 atoms) than those included in the training data set. PMID:28507695

Comparison of deep neural networks to spatio-temporal cortical dynamics of human visual object recognition reveals hierarchical correspondence

PubMed Central

Cichy, Radoslaw Martin; Khosla, Aditya; Pantazis, Dimitrios; Torralba, Antonio; Oliva, Aude

2016-01-01

The complex multi-stage architecture of cortical visual pathways provides the neural basis for efficient visual object recognition in humans. However, the stage-wise computations therein remain poorly understood. Here, we compared temporal (magnetoencephalography) and spatial (functional MRI) visual brain representations with representations in an artificial deep neural network (DNN) tuned to the statistics of real-world visual recognition. We showed that the DNN captured the stages of human visual processing in both time and space from early visual areas towards the dorsal and ventral streams. Further investigation of crucial DNN parameters revealed that while model architecture was important, training on real-world categorization was necessary to enforce spatio-temporal hierarchical relationships with the brain. Together our results provide an algorithmically informed view on the spatio-temporal dynamics of visual object recognition in the human visual brain. PMID:27282108

Comparison of deep neural networks to spatio-temporal cortical dynamics of human visual object recognition reveals hierarchical correspondence.

PubMed

Cichy, Radoslaw Martin; Khosla, Aditya; Pantazis, Dimitrios; Torralba, Antonio; Oliva, Aude

2016-06-10

The complex multi-stage architecture of cortical visual pathways provides the neural basis for efficient visual object recognition in humans. However, the stage-wise computations therein remain poorly understood. Here, we compared temporal (magnetoencephalography) and spatial (functional MRI) visual brain representations with representations in an artificial deep neural network (DNN) tuned to the statistics of real-world visual recognition. We showed that the DNN captured the stages of human visual processing in both time and space from early visual areas towards the dorsal and ventral streams. Further investigation of crucial DNN parameters revealed that while model architecture was important, training on real-world categorization was necessary to enforce spatio-temporal hierarchical relationships with the brain. Together our results provide an algorithmically informed view on the spatio-temporal dynamics of visual object recognition in the human visual brain.

Convolutional Neural Networks for 1-D Many-Channel Data

DTIC Science & Technology

Deep convolutional neural networks (CNNs) represent the state of the art in image recognition. The same properties that led to their success in that... crack detection ( 8,000 data points, 72 channels). Though the models predictive ability is limited to fitting the trend , its partial success suggests that...originally written to classify digits in the MNIST database (28 28 pixels, 1 channel), for use on 1-D acoustic data taken from experiments focused on

Residual Shuffling Convolutional Neural Networks for Deep Semantic Image Segmentation Using Multi-Modal Data

NASA Astrophysics Data System (ADS)

Chen, K.; Weinmann, M.; Gao, X.; Yan, M.; Hinz, S.; Jutzi, B.; Weinmann, M.

2018-05-01

In this paper, we address the deep semantic segmentation of aerial imagery based on multi-modal data. Given multi-modal data composed of true orthophotos and the corresponding Digital Surface Models (DSMs), we extract a variety of hand-crafted radiometric and geometric features which are provided separately and in different combinations as input to a modern deep learning framework. The latter is represented by a Residual Shuffling Convolutional Neural Network (RSCNN) combining the characteristics of a Residual Network with the advantages of atrous convolution and a shuffling operator to achieve a dense semantic labeling. Via performance evaluation on a benchmark dataset, we analyze the value of different feature sets for the semantic segmentation task. The derived results reveal that the use of radiometric features yields better classification results than the use of geometric features for the considered dataset. Furthermore, the consideration of data on both modalities leads to an improvement of the classification results. However, the derived results also indicate that the use of all defined features is less favorable than the use of selected features. Consequently, data representations derived via feature extraction and feature selection techniques still provide a gain if used as the basis for deep semantic segmentation.

Stellar Atmospheric Parameterization Based on Deep Learning

NASA Astrophysics Data System (ADS)

Pan, Ru-yang; Li, Xiang-ru

2017-07-01

Deep learning is a typical learning method widely studied in the fields of machine learning, pattern recognition, and artificial intelligence. This work investigates the problem of stellar atmospheric parameterization by constructing a deep neural network with five layers, and the node number in each layer of the network is respectively 3821-500-100-50-1. The proposed scheme is verified on both the real spectra measured by the Sloan Digital Sky Survey (SDSS) and the theoretic spectra computed with the Kurucz's New Opacity Distribution Function (NEWODF) model, to make an automatic estimation for three physical parameters: the effective temperature (Teff), surface gravitational acceleration (lg g), and metallic abundance (Fe/H). The results show that the stacked autoencoder deep neural network has a better accuracy for the estimation. On the SDSS spectra, the mean absolute errors (MAEs) are 79.95 for Teff/K, 0.0058 for (lg Teff/K), 0.1706 for lg (g/(cm·s-2)), and 0.1294 dex for the [Fe/H], respectively; On the theoretic spectra, the MAEs are 15.34 for Teff/K, 0.0011 for lg (Teff/K), 0.0214 for lg(g/(cm · s-2)), and 0.0121 dex for [Fe/H], respectively.

Rock images classification by using deep convolution neural network

NASA Astrophysics Data System (ADS)

Cheng, Guojian; Guo, Wenhui

2017-08-01

Granularity analysis is one of the most essential issues in authenticate under microscope. To improve the efficiency and accuracy of traditional manual work, an convolutional neural network based method is proposed for granularity analysis from thin section image, which chooses and extracts features from image samples while build classifier to recognize granularity of input image samples. 4800 samples from Ordos basin are used for experiments under colour spaces of HSV, YCbCr and RGB respectively. On the test dataset, the correct rate in RGB colour space is 98.5%, and it is believable in HSV and YCbCr colour space. The results show that the convolution neural network can classify the rock images with high reliability.

Segmentation and classification of colon glands with deep convolutional neural networks and total variation regularization.

PubMed

Kainz, Philipp; Pfeiffer, Michael; Urschler, Martin

2017-01-01

Segmentation of histopathology sections is a necessary preprocessing step for digital pathology. Due to the large variability of biological tissue, machine learning techniques have shown superior performance over conventional image processing methods. Here we present our deep neural network-based approach for segmentation and classification of glands in tissue of benign and malignant colorectal cancer, which was developed to participate in the GlaS@MICCAI2015 colon gland segmentation challenge. We use two distinct deep convolutional neural networks (CNN) for pixel-wise classification of Hematoxylin-Eosin stained images. While the first classifier separates glands from background, the second classifier identifies gland-separating structures. In a subsequent step, a figure-ground segmentation based on weighted total variation produces the final segmentation result by regularizing the CNN predictions. We present both quantitative and qualitative segmentation results on the recently released and publicly available Warwick-QU colon adenocarcinoma dataset associated with the GlaS@MICCAI2015 challenge and compare our approach to the simultaneously developed other approaches that participated in the same challenge. On two test sets, we demonstrate our segmentation performance and show that we achieve a tissue classification accuracy of 98% and 95%, making use of the inherent capability of our system to distinguish between benign and malignant tissue. Our results show that deep learning approaches can yield highly accurate and reproducible results for biomedical image analysis, with the potential to significantly improve the quality and speed of medical diagnoses.

Segmentation and classification of colon glands with deep convolutional neural networks and total variation regularization

PubMed Central

Kainz, Philipp; Pfeiffer, Michael

2017-01-01

Segmentation of histopathology sections is a necessary preprocessing step for digital pathology. Due to the large variability of biological tissue, machine learning techniques have shown superior performance over conventional image processing methods. Here we present our deep neural network-based approach for segmentation and classification of glands in tissue of benign and malignant colorectal cancer, which was developed to participate in the GlaS@MICCAI2015 colon gland segmentation challenge. We use two distinct deep convolutional neural networks (CNN) for pixel-wise classification of Hematoxylin-Eosin stained images. While the first classifier separates glands from background, the second classifier identifies gland-separating structures. In a subsequent step, a figure-ground segmentation based on weighted total variation produces the final segmentation result by regularizing the CNN predictions. We present both quantitative and qualitative segmentation results on the recently released and publicly available Warwick-QU colon adenocarcinoma dataset associated with the GlaS@MICCAI2015 challenge and compare our approach to the simultaneously developed other approaches that participated in the same challenge. On two test sets, we demonstrate our segmentation performance and show that we achieve a tissue classification accuracy of 98% and 95%, making use of the inherent capability of our system to distinguish between benign and malignant tissue. Our results show that deep learning approaches can yield highly accurate and reproducible results for biomedical image analysis, with the potential to significantly improve the quality and speed of medical diagnoses. PMID:29018612

Quantum-chemical insights from deep tensor neural networks

PubMed Central

Schütt, Kristof T.; Arbabzadah, Farhad; Chmiela, Stefan; Müller, Klaus R.; Tkatchenko, Alexandre

2017-01-01

Learning from data has led to paradigm shifts in a multitude of disciplines, including web, text and image search, speech recognition, as well as bioinformatics. Can machine learning enable similar breakthroughs in understanding quantum many-body systems? Here we develop an efficient deep learning approach that enables spatially and chemically resolved insights into quantum-mechanical observables of molecular systems. We unify concepts from many-body Hamiltonians with purpose-designed deep tensor neural networks, which leads to size-extensive and uniformly accurate (1 kcal mol−1) predictions in compositional and configurational chemical space for molecules of intermediate size. As an example of chemical relevance, the model reveals a classification of aromatic rings with respect to their stability. Further applications of our model for predicting atomic energies and local chemical potentials in molecules, reliable isomer energies, and molecules with peculiar electronic structure demonstrate the potential of machine learning for revealing insights into complex quantum-chemical systems. PMID:28067221

Quantum-chemical insights from deep tensor neural networks.

PubMed

Schütt, Kristof T; Arbabzadah, Farhad; Chmiela, Stefan; Müller, Klaus R; Tkatchenko, Alexandre

2017-01-09

Learning from data has led to paradigm shifts in a multitude of disciplines, including web, text and image search, speech recognition, as well as bioinformatics. Can machine learning enable similar breakthroughs in understanding quantum many-body systems? Here we develop an efficient deep learning approach that enables spatially and chemically resolved insights into quantum-mechanical observables of molecular systems. We unify concepts from many-body Hamiltonians with purpose-designed deep tensor neural networks, which leads to size-extensive and uniformly accurate (1 kcal mol -1 ) predictions in compositional and configurational chemical space for molecules of intermediate size. As an example of chemical relevance, the model reveals a classification of aromatic rings with respect to their stability. Further applications of our model for predicting atomic energies and local chemical potentials in molecules, reliable isomer energies, and molecules with peculiar electronic structure demonstrate the potential of machine learning for revealing insights into complex quantum-chemical systems.

Low Data Drug Discovery with One-Shot Learning.

PubMed

Altae-Tran, Han; Ramsundar, Bharath; Pappu, Aneesh S; Pande, Vijay

2017-04-26

Recent advances in machine learning have made significant contributions to drug discovery. Deep neural networks in particular have been demonstrated to provide significant boosts in predictive power when inferring the properties and activities of small-molecule compounds (Ma, J. et al. J. Chem. Inf. 2015, 55, 263-274). However, the applicability of these techniques has been limited by the requirement for large amounts of training data. In this work, we demonstrate how one-shot learning can be used to significantly lower the amounts of data required to make meaningful predictions in drug discovery applications. We introduce a new architecture, the iterative refinement long short-term memory, that, when combined with graph convolutional neural networks, significantly improves learning of meaningful distance metrics over small-molecules. We open source all models introduced in this work as part of DeepChem, an open-source framework for deep-learning in drug discovery (Ramsundar, B. deepchem.io. https://github.com/deepchem/deepchem, 2016).

Quantum-chemical insights from deep tensor neural networks

NASA Astrophysics Data System (ADS)

Schütt, Kristof T.; Arbabzadah, Farhad; Chmiela, Stefan; Müller, Klaus R.; Tkatchenko, Alexandre

2017-01-01

Learning from data has led to paradigm shifts in a multitude of disciplines, including web, text and image search, speech recognition, as well as bioinformatics. Can machine learning enable similar breakthroughs in understanding quantum many-body systems? Here we develop an efficient deep learning approach that enables spatially and chemically resolved insights into quantum-mechanical observables of molecular systems. We unify concepts from many-body Hamiltonians with purpose-designed deep tensor neural networks, which leads to size-extensive and uniformly accurate (1 kcal mol-1) predictions in compositional and configurational chemical space for molecules of intermediate size. As an example of chemical relevance, the model reveals a classification of aromatic rings with respect to their stability. Further applications of our model for predicting atomic energies and local chemical potentials in molecules, reliable isomer energies, and molecules with peculiar electronic structure demonstrate the potential of machine learning for revealing insights into complex quantum-chemical systems.

A Deep Learning Approach to LIBS Spectroscopy for Planetary Applications

NASA Astrophysics Data System (ADS)

Mullen, T. H.; Parente, M.; Gemp, I.; Dyar, M. D.

2017-12-01

The ChemCam instrument on the Curiousity rover has collected >440,000 laser-induced breakdown spectra (LIBS) from 1500 different geological targets since 2012. The team is using a pipeline of preprocessing and partial least squares techniques to predict compositions of surface materials [1]. Unfortunately, such multivariate techniques are plagued by hard-to-meet assumptions involving constant hyperparameter tuning to specific elements and the amount of training data available; if the whole distribution of data is not seen, the method will overfit to the training data and generalizability will suffer. The rover only has 10 calibration targets on-board that represent a small subset of the geochemical samples the rover is expected to investigate. Deep neural networks have been used to bypass these issues in other fields. Semi-supervised techniques allow researchers to utilized small labeled datasets and vast amounts of unlabeled data. One example is the variational autoencoder model, a semi-supervised generative model in the form of a deep neural network. The autoencoder assumes that LIBS spectra are generated from a distribution conditioned on the elemental compositions in the sample and some nuisance. The system is broken into two models: one that predicts elemental composition from the spectra and one that generates spectra from compositions that may or may not be seen in the training set. The synthesized spectra show strong agreement with geochemical conventions to express specific compositions. The predictions of composition show improved generalizability to PLS. Deep neural networks have also been used to transfer knowledge from one dataset to another to solve unlabeled data problems. Given that vast amounts of laboratry LIBS spectra have been obtained in the past few years, it is now feasible train a deep net to predict elemental composition from lab spectra. Transfer learning (manifold alignment or calibration transfer) [2] is then used to fine-tune the model from terrestrial lab data to Martian field data. Neural networks and generative models provide the flexibility need for elemental composition prediction and unseen spectra synthesis. [1] Clegg S. et al. (2016) Spectrochim. Acta B, 129, 64-85. [2] Boucher T. et al. (2017) J. Chemom., 31, e2877.

Deep Convolutional Framelet Denosing for Low-Dose CT via Wavelet Residual Network.

PubMed

Kang, Eunhee; Chang, Won; Yoo, Jaejun; Ye, Jong Chul

2018-06-01

Model-based iterative reconstruction algorithms for low-dose X-ray computed tomography (CT) are computationally expensive. To address this problem, we recently proposed a deep convolutional neural network (CNN) for low-dose X-ray CT and won the second place in 2016 AAPM Low-Dose CT Grand Challenge. However, some of the textures were not fully recovered. To address this problem, here we propose a novel framelet-based denoising algorithm using wavelet residual network which synergistically combines the expressive power of deep learning and the performance guarantee from the framelet-based denoising algorithms. The new algorithms were inspired by the recent interpretation of the deep CNN as a cascaded convolution framelet signal representation. Extensive experimental results confirm that the proposed networks have significantly improved performance and preserve the detail texture of the original images.

Mastering the game of Go with deep neural networks and tree search

NASA Astrophysics Data System (ADS)

Silver, David; Huang, Aja; Maddison, Chris J.; Guez, Arthur; Sifre, Laurent; van den Driessche, George; Schrittwieser, Julian; Antonoglou, Ioannis; Panneershelvam, Veda; Lanctot, Marc; Dieleman, Sander; Grewe, Dominik; Nham, John; Kalchbrenner, Nal; Sutskever, Ilya; Lillicrap, Timothy; Leach, Madeleine; Kavukcuoglu, Koray; Graepel, Thore; Hassabis, Demis

2016-01-01

The game of Go has long been viewed as the most challenging of classic games for artificial intelligence owing to its enormous search space and the difficulty of evaluating board positions and moves. Here we introduce a new approach to computer Go that uses ‘value networks’ to evaluate board positions and ‘policy networks’ to select moves. These deep neural networks are trained by a novel combination of supervised learning from human expert games, and reinforcement learning from games of self-play. Without any lookahead search, the neural networks play Go at the level of state-of-the-art Monte Carlo tree search programs that simulate thousands of random games of self-play. We also introduce a new search algorithm that combines Monte Carlo simulation with value and policy networks. Using this search algorithm, our program AlphaGo achieved a 99.8% winning rate against other Go programs, and defeated the human European Go champion by 5 games to 0. This is the first time that a computer program has defeated a human professional player in the full-sized game of Go, a feat previously thought to be at least a decade away.

Characterizing Deep Brain Stimulation effects in computationally efficient neural network models.

PubMed

Latteri, Alberta; Arena, Paolo; Mazzone, Paolo

2011-04-15

Recent studies on the medical treatment of Parkinson's disease (PD) led to the introduction of the so called Deep Brain Stimulation (DBS) technique. This particular therapy allows to contrast actively the pathological activity of various Deep Brain structures, responsible for the well known PD symptoms. This technique, frequently joined to dopaminergic drugs administration, replaces the surgical interventions implemented to contrast the activity of specific brain nuclei, called Basal Ganglia (BG). This clinical protocol gave the possibility to analyse and inspect signals measured from the electrodes implanted into the deep brain regions. The analysis of these signals led to the possibility to study the PD as a specific case of dynamical synchronization in biological neural networks, with the advantage to apply the theoretical analysis developed in such scientific field to find efficient treatments to face with this important disease. Experimental results in fact show that the PD neurological diseases are characterized by a pathological signal synchronization in BG. Parkinsonian tremor, for example, is ascribed to be caused by neuron populations of the Thalamic and Striatal structures that undergo an abnormal synchronization. On the contrary, in normal conditions, the activity of the same neuron populations do not appear to be correlated and synchronized. To study in details the effect of the stimulation signal on a pathological neural medium, efficient models of these neural structures were built, which are able to show, without any external input, the intrinsic properties of a pathological neural tissue, mimicking the BG synchronized dynamics.We start considering a model already introduced in the literature to investigate the effects of electrical stimulation on pathologically synchronized clusters of neurons. This model used Morris Lecar type neurons. This neuron model, although having a high level of biological plausibility, requires a large computational effort to simulate large scale networks. For this reason we considered a reduced order model, the Izhikevich one, which is computationally much lighter. The comparison between neural lattices built using both neuron models provided comparable results, both without traditional stimulation and in presence of all the stimulation protocols. This was a first result toward the study and simulation of the large scale neural networks involved in pathological dynamics.Using the reduced order model an inspection on the activity of two neural lattices was also carried out at the aim to analyze how the stimulation in one area could affect the dynamics in another area, like the usual medical treatment protocols require.The study of population dynamics that was carried out allowed us to investigate, through simulations, the positive effects of the stimulation signals in terms of desynchronization of the neural dynamics. The results obtained constitute a significant added value to the analysis of synchronization and desynchronization effects due to neural stimulation. This work gives the opportunity to more efficiently study the effect of stimulation in large scale yet computationally efficient neural networks. Results were compared both with the other mathematical models, using Morris Lecar and Izhikevich neurons, and with simulated Local Field Potentials (LFP).

A computational framework for the detection of subcortical brain dysmaturation in neonatal MRI using 3D Convolutional Neural Networks.

PubMed

Ceschin, Rafael; Zahner, Alexandria; Reynolds, William; Gaesser, Jenna; Zuccoli, Giulio; Lo, Cecilia W; Gopalakrishnan, Vanathi; Panigrahy, Ashok

2018-05-21

Deep neural networks are increasingly being used in both supervised learning for classification tasks and unsupervised learning to derive complex patterns from the input data. However, the successful implementation of deep neural networks using neuroimaging datasets requires adequate sample size for training and well-defined signal intensity based structural differentiation. There is a lack of effective automated diagnostic tools for the reliable detection of brain dysmaturation in the neonatal period, related to small sample size and complex undifferentiated brain structures, despite both translational research and clinical importance. Volumetric information alone is insufficient for diagnosis. In this study, we developed a computational framework for the automated classification of brain dysmaturation from neonatal MRI, by combining a specific deep neural network implementation with neonatal structural brain segmentation as a method for both clinical pattern recognition and data-driven inference into the underlying structural morphology. We implemented three-dimensional convolution neural networks (3D-CNNs) to specifically classify dysplastic cerebelli, a subset of surface-based subcortical brain dysmaturation, in term infants born with congenital heart disease. We obtained a 0.985 ± 0. 0241-classification accuracy of subtle cerebellar dysplasia in CHD using 10-fold cross-validation. Furthermore, the hidden layer activations and class activation maps depicted regional vulnerability of the superior surface of the cerebellum, (composed of mostly the posterior lobe and the midline vermis), in regards to differentiating the dysplastic process from normal tissue. The posterior lobe and the midline vermis provide regional differentiation that is relevant to not only to the clinical diagnosis of cerebellar dysplasia, but also genetic mechanisms and neurodevelopmental outcome correlates. These findings not only contribute to the detection and classification of a subset of neonatal brain dysmaturation, but also provide insight to the pathogenesis of cerebellar dysplasia in CHD. In addition, this is one of the first examples of the application of deep learning to a neuroimaging dataset, in which the hidden layer activation revealed diagnostically and biologically relevant features about the clinical pathogenesis. The code developed for this project is open source, published under the BSD License, and designed to be generalizable to applications both within and beyond neonatal brain imaging. Copyright © 2018 Elsevier Inc. All rights reserved.

Using Deep Learning for Targeted Data Selection, Improving Satellite Observation Utilization for Model Initialization

NASA Astrophysics Data System (ADS)

Lee, Y. J.; Bonfanti, C. E.; Trailovic, L.; Etherton, B.; Govett, M.; Stewart, J.

2017-12-01

At present, a fraction of all satellite observations are ultimately used for model assimilation. The satellite data assimilation process is computationally expensive and data are often reduced in resolution to allow timely incorporation into the forecast. This problem is only exacerbated by the recent launch of Geostationary Operational Environmental Satellite (GOES)-16 satellite and future satellites providing several order of magnitude increase in data volume. At the NOAA Earth System Research Laboratory (ESRL) we are researching the use of machine learning the improve the initial selection of satellite data to be used in the model assimilation process. In particular, we are investigating the use of deep learning. Deep learning is being applied to many image processing and computer vision problems with great success. Through our research, we are using convolutional neural network to find and mark regions of interest (ROI) to lead to intelligent extraction of observations from satellite observation systems. These targeted observations will be used to improve the quality of data selected for model assimilation and ultimately improve the impact of satellite data on weather forecasts. Our preliminary efforts to identify the ROI's are focused in two areas: applying and comparing state-of-art convolutional neural network models using the analysis data from the National Center for Environmental Prediction (NCEP) Global Forecast System (GFS) weather model, and using these results as a starting point to optimize convolution neural network model for pattern recognition on the higher resolution water vapor data from GOES-WEST and other satellite. This presentation will provide an introduction to our convolutional neural network model to identify and process these ROI's, along with the challenges of data preparation, training the model, and parameter optimization.

Weed Growth Stage Estimator Using Deep Convolutional Neural Networks.

PubMed

Teimouri, Nima; Dyrmann, Mads; Nielsen, Per Rydahl; Mathiassen, Solvejg Kopp; Somerville, Gayle J; Jørgensen, Rasmus Nyholm

2018-05-16

This study outlines a new method of automatically estimating weed species and growth stages (from cotyledon until eight leaves are visible) of in situ images covering 18 weed species or families. Images of weeds growing within a variety of crops were gathered across variable environmental conditions with regards to soil types, resolution and light settings. Then, 9649 of these images were used for training the computer, which automatically divided the weeds into nine growth classes. The performance of this proposed convolutional neural network approach was evaluated on a further set of 2516 images, which also varied in term of crop, soil type, image resolution and light conditions. The overall performance of this approach achieved a maximum accuracy of 78% for identifying Polygonum spp. and a minimum accuracy of 46% for blackgrass. In addition, it achieved an average 70% accuracy rate in estimating the number of leaves and 96% accuracy when accepting a deviation of two leaves. These results show that this new method of using deep convolutional neural networks has a relatively high ability to estimate early growth stages across a wide variety of weed species.

Reinforced dynamics for enhanced sampling in large atomic and molecular systems

NASA Astrophysics Data System (ADS)

Zhang, Linfeng; Wang, Han; E, Weinan

2018-03-01

A new approach for efficiently exploring the configuration space and computing the free energy of large atomic and molecular systems is proposed, motivated by an analogy with reinforcement learning. There are two major components in this new approach. Like metadynamics, it allows for an efficient exploration of the configuration space by adding an adaptively computed biasing potential to the original dynamics. Like deep reinforcement learning, this biasing potential is trained on the fly using deep neural networks, with data collected judiciously from the exploration and an uncertainty indicator from the neural network model playing the role of the reward function. Parameterization using neural networks makes it feasible to handle cases with a large set of collective variables. This has the potential advantage that selecting precisely the right set of collective variables has now become less critical for capturing the structural transformations of the system. The method is illustrated by studying the full-atom explicit solvent models of alanine dipeptide and tripeptide, as well as the system of a polyalanine-10 molecule with 20 collective variables.

Fusion of shallow and deep features for classification of high-resolution remote sensing images

NASA Astrophysics Data System (ADS)

Gao, Lang; Tian, Tian; Sun, Xiao; Li, Hang

2018-02-01

Effective spectral and spatial pixel description plays a significant role for the classification of high resolution remote sensing images. Current approaches of pixel-based feature extraction are of two main kinds: one includes the widelyused principal component analysis (PCA) and gray level co-occurrence matrix (GLCM) as the representative of the shallow spectral and shape features, and the other refers to the deep learning-based methods which employ deep neural networks and have made great promotion on classification accuracy. However, the former traditional features are insufficient to depict complex distribution of high resolution images, while the deep features demand plenty of samples to train the network otherwise over fitting easily occurs if only limited samples are involved in the training. In view of the above, we propose a GLCM-based convolution neural network (CNN) approach to extract features and implement classification for high resolution remote sensing images. The employment of GLCM is able to represent the original images and eliminate redundant information and undesired noises. Meanwhile, taking shallow features as the input of deep network will contribute to a better guidance and interpretability. In consideration of the amount of samples, some strategies such as L2 regularization and dropout methods are used to prevent over-fitting. The fine-tuning strategy is also used in our study to reduce training time and further enhance the generalization performance of the network. Experiments with popular data sets such as PaviaU data validate that our proposed method leads to a performance improvement compared to individual involved approaches.

What the success of brain imaging implies about the neural code.

PubMed

Guest, Olivia; Love, Bradley C

2017-01-19

The success of fMRI places constraints on the nature of the neural code. The fact that researchers can infer similarities between neural representations, despite fMRI's limitations, implies that certain neural coding schemes are more likely than others. For fMRI to succeed given its low temporal and spatial resolution, the neural code must be smooth at the voxel and functional level such that similar stimuli engender similar internal representations. Through proof and simulation, we determine which coding schemes are plausible given both fMRI's successes and its limitations in measuring neural activity. Deep neural network approaches, which have been forwarded as computational accounts of the ventral stream, are consistent with the success of fMRI, though functional smoothness breaks down in the later network layers. These results have implications for the nature of the neural code and ventral stream, as well as what can be successfully investigated with fMRI.

Detection and Localization of Robotic Tools in Robot-Assisted Surgery Videos Using Deep Neural Networks for Region Proposal and Detection.

PubMed

Sarikaya, Duygu; Corso, Jason J; Guru, Khurshid A

2017-07-01

Video understanding of robot-assisted surgery (RAS) videos is an active research area. Modeling the gestures and skill level of surgeons presents an interesting problem. The insights drawn may be applied in effective skill acquisition, objective skill assessment, real-time feedback, and human-robot collaborative surgeries. We propose a solution to the tool detection and localization open problem in RAS video understanding, using a strictly computer vision approach and the recent advances of deep learning. We propose an architecture using multimodal convolutional neural networks for fast detection and localization of tools in RAS videos. To the best of our knowledge, this approach will be the first to incorporate deep neural networks for tool detection and localization in RAS videos. Our architecture applies a region proposal network (RPN) and a multimodal two stream convolutional network for object detection to jointly predict objectness and localization on a fusion of image and temporal motion cues. Our results with an average precision of 91% and a mean computation time of 0.1 s per test frame detection indicate that our study is superior to conventionally used methods for medical imaging while also emphasizing the benefits of using RPN for precision and efficiency. We also introduce a new data set, ATLAS Dione, for RAS video understanding. Our data set provides video data of ten surgeons from Roswell Park Cancer Institute, Buffalo, NY, USA, performing six different surgical tasks on the daVinci Surgical System (dVSS) with annotations of robotic tools per frame.

Automatic Detection of Welding Defects using Deep Neural Network

NASA Astrophysics Data System (ADS)

Hou, Wenhui; Wei, Ye; Guo, Jie; Jin, Yi; Zhu, Chang'an

2018-01-01

In this paper, we propose an automatic detection schema including three stages for weld defects in x-ray images. Firstly, the preprocessing procedure for the image is implemented to locate the weld region; Then a classification model which is trained and tested by the patches cropped from x-ray images is constructed based on deep neural network. And this model can learn the intrinsic feature of images without extra calculation; Finally, the sliding-window approach is utilized to detect the whole images based on the trained model. In order to evaluate the performance of the model, we carry out several experiments. The results demonstrate that the classification model we proposed is effective in the detection of welded joints quality.

Deep Learning for Magnetic Resonance Fingerprinting: A New Approach for Predicting Quantitative Parameter Values from Time Series.

PubMed

Hoppe, Elisabeth; Körzdörfer, Gregor; Würfl, Tobias; Wetzl, Jens; Lugauer, Felix; Pfeuffer, Josef; Maier, Andreas

2017-01-01

The purpose of this work is to evaluate methods from deep learning for application to Magnetic Resonance Fingerprinting (MRF). MRF is a recently proposed measurement technique for generating quantitative parameter maps. In MRF a non-steady state signal is generated by a pseudo-random excitation pattern. A comparison of the measured signal in each voxel with the physical model yields quantitative parameter maps. Currently, the comparison is done by matching a dictionary of simulated signals to the acquired signals. To accelerate the computation of quantitative maps we train a Convolutional Neural Network (CNN) on simulated dictionary data. As a proof of principle we show that the neural network implicitly encodes the dictionary and can replace the matching process.

Lifelong learning of human actions with deep neural network self-organization.

PubMed

Parisi, German I; Tani, Jun; Weber, Cornelius; Wermter, Stefan

2017-12-01

Lifelong learning is fundamental in autonomous robotics for the acquisition and fine-tuning of knowledge through experience. However, conventional deep neural models for action recognition from videos do not account for lifelong learning but rather learn a batch of training data with a predefined number of action classes and samples. Thus, there is the need to develop learning systems with the ability to incrementally process available perceptual cues and to adapt their responses over time. We propose a self-organizing neural architecture for incrementally learning to classify human actions from video sequences. The architecture comprises growing self-organizing networks equipped with recurrent neurons for processing time-varying patterns. We use a set of hierarchically arranged recurrent networks for the unsupervised learning of action representations with increasingly large spatiotemporal receptive fields. Lifelong learning is achieved in terms of prediction-driven neural dynamics in which the growth and the adaptation of the recurrent networks are driven by their capability to reconstruct temporally ordered input sequences. Experimental results on a classification task using two action benchmark datasets show that our model is competitive with state-of-the-art methods for batch learning also when a significant number of sample labels are missing or corrupted during training sessions. Additional experiments show the ability of our model to adapt to non-stationary input avoiding catastrophic interference. Copyright © 2017 The Author(s). Published by Elsevier Ltd.. All rights reserved.

Can surgical simulation be used to train detection and classification of neural networks?

PubMed

Zisimopoulos, Odysseas; Flouty, Evangello; Stacey, Mark; Muscroft, Sam; Giataganas, Petros; Nehme, Jean; Chow, Andre; Stoyanov, Danail

2017-10-01

Computer-assisted interventions (CAI) aim to increase the effectiveness, precision and repeatability of procedures to improve surgical outcomes. The presence and motion of surgical tools is a key information input for CAI surgical phase recognition algorithms. Vision-based tool detection and recognition approaches are an attractive solution and can be designed to take advantage of the powerful deep learning paradigm that is rapidly advancing image recognition and classification. The challenge for such algorithms is the availability and quality of labelled data used for training. In this Letter, surgical simulation is used to train tool detection and segmentation based on deep convolutional neural networks and generative adversarial networks. The authors experiment with two network architectures for image segmentation in tool classes commonly encountered during cataract surgery. A commercially-available simulator is used to create a simulated cataract dataset for training models prior to performing transfer learning on real surgical data. To the best of authors' knowledge, this is the first attempt to train deep learning models for surgical instrument detection on simulated data while demonstrating promising results to generalise on real data. Results indicate that simulated data does have some potential for training advanced classification methods for CAI systems.

Interaction of oscillations, and their suppression via deep brain stimulation, in a model of the cortico-basal ganglia network.

PubMed

Kang, Guiyeom; Lowery, Madeleine M

2013-03-01

Growing evidence suggests that synchronized neural oscillations in the cortico-basal ganglia network may play a critical role in the pathophysiology of Parkinson's disease. In this study, a new model of the closed loop network is used to explore the generation and interaction of network oscillations and their suppression through deep brain stimulation (DBS). Under simulated dopamine depletion conditions, increased gain through the hyperdirect pathway resulted in the interaction of neural oscillations at different frequencies in the cortex and subthalamic nucleus (STN), leading to the emergence of synchronized oscillations at a new intermediate frequency. Further increases in synaptic gain resulted in the cortex driving synchronous oscillatory activity throughout the network. When DBS was added to the model a progressive reduction in STN power at the tremor and beta frequencies was observed as the frequency of stimulation was increased, with resonance effects occurring for low frequency DBS (40 Hz) in agreement with experimental observations. The results provide new insights into the mechanisms by which synchronous oscillations can arise within the network and how DBS may suppress unwanted oscillatory activity.

Deep Spatial-Temporal Joint Feature Representation for Video Object Detection.

PubMed

Zhao, Baojun; Zhao, Boya; Tang, Linbo; Han, Yuqi; Wang, Wenzheng

2018-03-04

With the development of deep neural networks, many object detection frameworks have shown great success in the fields of smart surveillance, self-driving cars, and facial recognition. However, the data sources are usually videos, and the object detection frameworks are mostly established on still images and only use the spatial information, which means that the feature consistency cannot be ensured because the training procedure loses temporal information. To address these problems, we propose a single, fully-convolutional neural network-based object detection framework that involves temporal information by using Siamese networks. In the training procedure, first, the prediction network combines the multiscale feature map to handle objects of various sizes. Second, we introduce a correlation loss by using the Siamese network, which provides neighboring frame features. This correlation loss represents object co-occurrences across time to aid the consistent feature generation. Since the correlation loss should use the information of the track ID and detection label, our video object detection network has been evaluated on the large-scale ImageNet VID dataset where it achieves a 69.5% mean average precision (mAP).

Stochastic Spiking Neural Networks Enabled by Magnetic Tunnel Junctions: From Nontelegraphic to Telegraphic Switching Regimes

NASA Astrophysics Data System (ADS)

Liyanagedera, Chamika M.; Sengupta, Abhronil; Jaiswal, Akhilesh; Roy, Kaushik

2017-12-01

Stochastic spiking neural networks based on nanoelectronic spin devices can be a possible pathway to achieving "brainlike" compact and energy-efficient cognitive intelligence. The computational model attempt to exploit the intrinsic device stochasticity of nanoelectronic synaptic or neural components to perform learning or inference. However, there has been limited analysis on the scaling effect of stochastic spin devices and its impact on the operation of such stochastic networks at the system level. This work attempts to explore the design space and analyze the performance of nanomagnet-based stochastic neuromorphic computing architectures for magnets with different barrier heights. We illustrate how the underlying network architecture must be modified to account for the random telegraphic switching behavior displayed by magnets with low barrier heights as they are scaled into the superparamagnetic regime. We perform a device-to-system-level analysis on a deep neural-network architecture for a digit-recognition problem on the MNIST data set.

DRREP: deep ridge regressed epitope predictor.

PubMed

Sher, Gene; Zhi, Degui; Zhang, Shaojie

2017-10-03

The ability to predict epitopes plays an enormous role in vaccine development in terms of our ability to zero in on where to do a more thorough in-vivo analysis of the protein in question. Though for the past decade there have been numerous advancements and improvements in epitope prediction, on average the best benchmark prediction accuracies are still only around 60%. New machine learning algorithms have arisen within the domain of deep learning, text mining, and convolutional networks. This paper presents a novel analytically trained and string kernel using deep neural network, which is tailored for continuous epitope prediction, called: Deep Ridge Regressed Epitope Predictor (DRREP). DRREP was tested on long protein sequences from the following datasets: SARS, Pellequer, HIV, AntiJen, and SEQ194. DRREP was compared to numerous state of the art epitope predictors, including the most recently published predictors called LBtope and DMNLBE. Using area under ROC curve (AUC), DRREP achieved a performance improvement over the best performing predictors on SARS (13.7%), HIV (8.9%), Pellequer (1.5%), and SEQ194 (3.1%), with its performance being matched only on the AntiJen dataset, by the LBtope predictor, where both DRREP and LBtope achieved an AUC of 0.702. DRREP is an analytically trained deep neural network, thus capable of learning in a single step through regression. By combining the features of deep learning, string kernels, and convolutional networks, the system is able to perform residue-by-residue prediction of continues epitopes with higher accuracy than the current state of the art predictors.

Classification of breast MRI lesions using small-size training sets: comparison of deep learning approaches

NASA Astrophysics Data System (ADS)

Amit, Guy; Ben-Ari, Rami; Hadad, Omer; Monovich, Einat; Granot, Noa; Hashoul, Sharbell

2017-03-01

Diagnostic interpretation of breast MRI studies requires meticulous work and a high level of expertise. Computerized algorithms can assist radiologists by automatically characterizing the detected lesions. Deep learning approaches have shown promising results in natural image classification, but their applicability to medical imaging is limited by the shortage of large annotated training sets. In this work, we address automatic classification of breast MRI lesions using two different deep learning approaches. We propose a novel image representation for dynamic contrast enhanced (DCE) breast MRI lesions, which combines the morphological and kinetics information in a single multi-channel image. We compare two classification approaches for discriminating between benign and malignant lesions: training a designated convolutional neural network and using a pre-trained deep network to extract features for a shallow classifier. The domain-specific trained network provided higher classification accuracy, compared to the pre-trained model, with an area under the ROC curve of 0.91 versus 0.81, and an accuracy of 0.83 versus 0.71. Similar accuracy was achieved in classifying benign lesions, malignant lesions, and normal tissue images. The trained network was able to improve accuracy by using the multi-channel image representation, and was more robust to reductions in the size of the training set. A small-size convolutional neural network can learn to accurately classify findings in medical images using only a few hundred images from a few dozen patients. With sufficient data augmentation, such a network can be trained to outperform a pre-trained out-of-domain classifier. Developing domain-specific deep-learning models for medical imaging can facilitate technological advancements in computer-aided diagnosis.

Mining e-cigarette adverse events in social media using Bi-LSTM recurrent neural network with word embedding representation.

PubMed

Xie, Jiaheng; Liu, Xiao; Dajun Zeng, Daniel

2018-01-01

Recent years have seen increased worldwide popularity of e-cigarette use. However, the risks of e-cigarettes are underexamined. Most e-cigarette adverse event studies have achieved low detection rates due to limited subject sample sizes in the experiments and surveys. Social media provides a large data repository of consumers' e-cigarette feedback and experiences, which are useful for e-cigarette safety surveillance. However, it is difficult to automatically interpret the informal and nontechnical consumer vocabulary about e-cigarettes in social media. This issue hinders the use of social media content for e-cigarette safety surveillance. Recent developments in deep neural network methods have shown promise for named entity extraction from noisy text. Motivated by these observations, we aimed to design a deep neural network approach to extract e-cigarette safety information in social media. Our deep neural language model utilizes word embedding as the representation of text input and recognizes named entity types with the state-of-the-art Bidirectional Long Short-Term Memory (Bi-LSTM) Recurrent Neural Network. Our Bi-LSTM model achieved the best performance compared to 3 baseline models, with a precision of 94.10%, a recall of 91.80%, and an F-measure of 92.94%. We identified 1591 unique adverse events and 9930 unique e-cigarette components (ie, chemicals, flavors, and devices) from our research testbed. Although the conditional random field baseline model had slightly better precision than our approach, our Bi-LSTM model achieved much higher recall, resulting in the best F-measure. Our method can be generalized to extract medical concepts from social media for other medical applications. © The Author 2017. Published by Oxford University Press on behalf of the American Medical Informatics Association. All rights reserved. For Permissions, please email: journals.permissions@oup.com

Deep nets vs expert designed features in medical physics: An IMRT QA case study.

PubMed

Interian, Yannet; Rideout, Vincent; Kearney, Vasant P; Gennatas, Efstathios; Morin, Olivier; Cheung, Joey; Solberg, Timothy; Valdes, Gilmer

2018-03-30

The purpose of this study was to compare the performance of Deep Neural Networks against a technique designed by domain experts in the prediction of gamma passing rates for Intensity Modulated Radiation Therapy Quality Assurance (IMRT QA). A total of 498 IMRT plans across all treatment sites were planned in Eclipse version 11 and delivered using a dynamic sliding window technique on Clinac iX or TrueBeam Linacs. Measurements were performed using a commercial 2D diode array, and passing rates for 3%/3 mm local dose/distance-to-agreement (DTA) were recorded. Separately, fluence maps calculated for each plan were used as inputs to a convolution neural network (CNN). The CNNs were trained to predict IMRT QA gamma passing rates using TensorFlow and Keras. A set of model architectures, inspired by the convolutional blocks of the VGG-16 ImageNet model, were constructed and implemented. Synthetic data, created by rotating and translating the fluence maps during training, was created to boost the performance of the CNNs. Dropout, batch normalization, and data augmentation were utilized to help train the model. The performance of the CNNs was compared to a generalized Poisson regression model, previously developed for this application, which used 78 expert designed features. Deep Neural Networks without domain knowledge achieved comparable performance to a baseline system designed by domain experts in the prediction of 3%/3 mm Local gamma passing rates. An ensemble of neural nets resulted in a mean absolute error (MAE) of 0.70 ± 0.05 and the domain expert model resulted in a 0.74 ± 0.06. Convolutional neural networks (CNNs) with transfer learning can predict IMRT QA passing rates by automatically designing features from the fluence maps without human expert supervision. Predictions from CNNs are comparable to a system carefully designed by physicist experts. © 2018 American Association of Physicists in Medicine.

Reinforced Adversarial Neural Computer for de Novo Molecular Design.

PubMed

Putin, Evgeny; Asadulaev, Arip; Ivanenkov, Yan; Aladinskiy, Vladimir; Sanchez-Lengeling, Benjamin; Aspuru-Guzik, Alán; Zhavoronkov, Alex

2018-06-12

In silico modeling is a crucial milestone in modern drug design and development. Although computer-aided approaches in this field are well-studied, the application of deep learning methods in this research area is at the beginning. In this work, we present an original deep neural network (DNN) architecture named RANC (Reinforced Adversarial Neural Computer) for the de novo design of novel small-molecule organic structures based on the generative adversarial network (GAN) paradigm and reinforcement learning (RL). As a generator RANC uses a differentiable neural computer (DNC), a category of neural networks, with increased generation capabilities due to the addition of an explicit memory bank, which can mitigate common problems found in adversarial settings. The comparative results have shown that RANC trained on the SMILES string representation of the molecules outperforms its first DNN-based counterpart ORGANIC by several metrics relevant to drug discovery: the number of unique structures, passing medicinal chemistry filters (MCFs), Muegge criteria, and high QED scores. RANC is able to generate structures that match the distributions of the key chemical features/descriptors (e.g., MW, logP, TPSA) and lengths of the SMILES strings in the training data set. Therefore, RANC can be reasonably regarded as a promising starting point to develop novel molecules with activity against different biological targets or pathways. In addition, this approach allows scientists to save time and covers a broad chemical space populated with novel and diverse compounds.

A Robust Deep-Learning-Based Detector for Real-Time Tomato Plant Diseases and Pests Recognition.

PubMed

Fuentes, Alvaro; Yoon, Sook; Kim, Sang Cheol; Park, Dong Sun

2017-09-04

Plant Diseases and Pests are a major challenge in the agriculture sector. An accurate and a faster detection of diseases and pests in plants could help to develop an early treatment technique while substantially reducing economic losses. Recent developments in Deep Neural Networks have allowed researchers to drastically improve the accuracy of object detection and recognition systems. In this paper, we present a deep-learning-based approach to detect diseases and pests in tomato plants using images captured in-place by camera devices with various resolutions. Our goal is to find the more suitable deep-learning architecture for our task. Therefore, we consider three main families of detectors: Faster Region-based Convolutional Neural Network (Faster R-CNN), Region-based Fully Convolutional Network (R-FCN), and Single Shot Multibox Detector (SSD), which for the purpose of this work are called "deep learning meta-architectures". We combine each of these meta-architectures with "deep feature extractors" such as VGG net and Residual Network (ResNet). We demonstrate the performance of deep meta-architectures and feature extractors, and additionally propose a method for local and global class annotation and data augmentation to increase the accuracy and reduce the number of false positives during training. We train and test our systems end-to-end on our large Tomato Diseases and Pests Dataset, which contains challenging images with diseases and pests, including several inter- and extra-class variations, such as infection status and location in the plant. Experimental results show that our proposed system can effectively recognize nine different types of diseases and pests, with the ability to deal with complex scenarios from a plant's surrounding area.

A Robust Deep-Learning-Based Detector for Real-Time Tomato Plant Diseases and Pests Recognition

PubMed Central

Yoon, Sook; Kim, Sang Cheol; Park, Dong Sun

2017-01-01

Plant Diseases and Pests are a major challenge in the agriculture sector. An accurate and a faster detection of diseases and pests in plants could help to develop an early treatment technique while substantially reducing economic losses. Recent developments in Deep Neural Networks have allowed researchers to drastically improve the accuracy of object detection and recognition systems. In this paper, we present a deep-learning-based approach to detect diseases and pests in tomato plants using images captured in-place by camera devices with various resolutions. Our goal is to find the more suitable deep-learning architecture for our task. Therefore, we consider three main families of detectors: Faster Region-based Convolutional Neural Network (Faster R-CNN), Region-based Fully Convolutional Network (R-FCN), and Single Shot Multibox Detector (SSD), which for the purpose of this work are called “deep learning meta-architectures”. We combine each of these meta-architectures with “deep feature extractors” such as VGG net and Residual Network (ResNet). We demonstrate the performance of deep meta-architectures and feature extractors, and additionally propose a method for local and global class annotation and data augmentation to increase the accuracy and reduce the number of false positives during training. We train and test our systems end-to-end on our large Tomato Diseases and Pests Dataset, which contains challenging images with diseases and pests, including several inter- and extra-class variations, such as infection status and location in the plant. Experimental results show that our proposed system can effectively recognize nine different types of diseases and pests, with the ability to deal with complex scenarios from a plant’s surrounding area. PMID:28869539

DeepX: Deep Learning Accelerator for Restricted Boltzmann Machine Artificial Neural Networks.

PubMed

Kim, Lok-Won

2018-05-01

Although there have been many decades of research and commercial presence on high performance general purpose processors, there are still many applications that require fully customized hardware architectures for further computational acceleration. Recently, deep learning has been successfully used to learn in a wide variety of applications, but their heavy computation demand has considerably limited their practical applications. This paper proposes a fully pipelined acceleration architecture to alleviate high computational demand of an artificial neural network (ANN) which is restricted Boltzmann machine (RBM) ANNs. The implemented RBM ANN accelerator (integrating network size, using 128 input cases per batch, and running at a 303-MHz clock frequency) integrated in a state-of-the art field-programmable gate array (FPGA) (Xilinx Virtex 7 XC7V-2000T) provides a computational performance of 301-billion connection-updates-per-second and about 193 times higher performance than a software solution running on general purpose processors. Most importantly, the architecture enables over 4 times (12 times in batch learning) higher performance compared with a previous work when both are implemented in an FPGA device (XC2VP70).

Deep Learning for Drug Design: an Artificial Intelligence Paradigm for Drug Discovery in the Big Data Era.

PubMed

Jing, Yankang; Bian, Yuemin; Hu, Ziheng; Wang, Lirong; Xie, Xiang-Qun Sean

2018-03-30

Over the last decade, deep learning (DL) methods have been extremely successful and widely used to develop artificial intelligence (AI) in almost every domain, especially after it achieved its proud record on computational Go. Compared to traditional machine learning (ML) algorithms, DL methods still have a long way to go to achieve recognition in small molecular drug discovery and development. And there is still lots of work to do for the popularization and application of DL for research purpose, e.g., for small molecule drug research and development. In this review, we mainly discussed several most powerful and mainstream architectures, including the convolutional neural network (CNN), recurrent neural network (RNN), and deep auto-encoder networks (DAENs), for supervised learning and nonsupervised learning; summarized most of the representative applications in small molecule drug design; and briefly introduced how DL methods were used in those applications. The discussion for the pros and cons of DL methods as well as the main challenges we need to tackle were also emphasized.

Deep neural networks: A promising tool for fault characteristic mining and intelligent diagnosis of rotating machinery with massive data

NASA Astrophysics Data System (ADS)

Jia, Feng; Lei, Yaguo; Lin, Jing; Zhou, Xin; Lu, Na

2016-05-01

Aiming to promptly process the massive fault data and automatically provide accurate diagnosis results, numerous studies have been conducted on intelligent fault diagnosis of rotating machinery. Among these studies, the methods based on artificial neural networks (ANNs) are commonly used, which employ signal processing techniques for extracting features and further input the features to ANNs for classifying faults. Though these methods did work in intelligent fault diagnosis of rotating machinery, they still have two deficiencies. (1) The features are manually extracted depending on much prior knowledge about signal processing techniques and diagnostic expertise. In addition, these manual features are extracted according to a specific diagnosis issue and probably unsuitable for other issues. (2) The ANNs adopted in these methods have shallow architectures, which limits the capacity of ANNs to learn the complex non-linear relationships in fault diagnosis issues. As a breakthrough in artificial intelligence, deep learning holds the potential to overcome the aforementioned deficiencies. Through deep learning, deep neural networks (DNNs) with deep architectures, instead of shallow ones, could be established to mine the useful information from raw data and approximate complex non-linear functions. Based on DNNs, a novel intelligent method is proposed in this paper to overcome the deficiencies of the aforementioned intelligent diagnosis methods. The effectiveness of the proposed method is validated using datasets from rolling element bearings and planetary gearboxes. These datasets contain massive measured signals involving different health conditions under various operating conditions. The diagnosis results show that the proposed method is able to not only adaptively mine available fault characteristics from the measured signals, but also obtain superior diagnosis accuracy compared with the existing methods.

High precision in protein contact prediction using fully convolutional neural networks and minimal sequence features.

PubMed

Jones, David T; Kandathil, Shaun M

2018-04-26

In addition to substitution frequency data from protein sequence alignments, many state-of-the-art methods for contact prediction rely on additional sources of information, or features, of protein sequences in order to predict residue-residue contacts, such as solvent accessibility, predicted secondary structure, and scores from other contact prediction methods. It is unclear how much of this information is needed to achieve state-of-the-art results. Here, we show that using deep neural network models, simple alignment statistics contain sufficient information to achieve state-of-the-art precision. Our prediction method, DeepCov, uses fully convolutional neural networks operating on amino-acid pair frequency or covariance data derived directly from sequence alignments, without using global statistical methods such as sparse inverse covariance or pseudolikelihood estimation. Comparisons against CCMpred and MetaPSICOV2 show that using pairwise covariance data calculated from raw alignments as input allows us to match or exceed the performance of both of these methods. Almost all of the achieved precision is obtained when considering relatively local windows (around 15 residues) around any member of a given residue pairing; larger window sizes have comparable performance. Assessment on a set of shallow sequence alignments (fewer than 160 effective sequences) indicates that the new method is substantially more precise than CCMpred and MetaPSICOV2 in this regime, suggesting that improved precision is attainable on smaller sequence families. Overall, the performance of DeepCov is competitive with the state of the art, and our results demonstrate that global models, which employ features from all parts of the input alignment when predicting individual contacts, are not strictly needed in order to attain precise contact predictions. DeepCov is freely available at https://github.com/psipred/DeepCov. d.t.jones@ucl.ac.uk.

Deep convolutional neural network for the classification of hepatocellular carcinoma and intrahepatic cholangiocarcinoma

NASA Astrophysics Data System (ADS)

Midya, Abhishek; Chakraborty, Jayasree; Pak, Linda M.; Zheng, Jian; Jarnagin, William R.; Do, Richard K. G.; Simpson, Amber L.

2018-02-01

Liver cancer is the second leading cause of cancer-related death worldwide.1 Hepatocellular carcinoma (HCC) is the most common primary liver cancer accounting for approximately 80% of cases. Intrahepatic cholangiocarcinoma (ICC) is a rare liver cancer, arising in patients with the same risk factors as HCC, but treatment options and prognosis differ. The diagnosis of HCC is based primarily on imaging but distinguishing between HCC and ICC is challenging due to common radiographic features.2-4 The aim of the present study is to classify HCC and ICC in portal venous phase CT. 107 patients with resected ICC and 116 patients with resected HCC were included in our analysis. We developed a deep neural network by modifying a pre-trained Inception network by retraining the final layers. The proposed method achieved the best accuracy and area under the receiver operating characteristics curve of 69.70% and 0.72, respectively on the test data.

Increasingly complex representations of natural movies across the dorsal stream are shared between subjects.

PubMed

Güçlü, Umut; van Gerven, Marcel A J

2017-01-15

Recently, deep neural networks (DNNs) have been shown to provide accurate predictions of neural responses across the ventral visual pathway. We here explore whether they also provide accurate predictions of neural responses across the dorsal visual pathway, which is thought to be devoted to motion processing and action recognition. This is achieved by training deep neural networks to recognize actions in videos and subsequently using them to predict neural responses while subjects are watching natural movies. Moreover, we explore whether dorsal stream representations are shared between subjects. In order to address this question, we examine if individual subject predictions can be made in a common representational space estimated via hyperalignment. Results show that a DNN trained for action recognition can be used to accurately predict how dorsal stream responds to natural movies, revealing a correspondence in representations of DNN layers and dorsal stream areas. It is also demonstrated that models operating in a common representational space can generalize to responses of multiple or even unseen individual subjects to novel spatio-temporal stimuli in both encoding and decoding settings, suggesting that a common representational space underlies dorsal stream responses across multiple subjects. Copyright © 2015 Elsevier Inc. All rights reserved.

Robust Vehicle Detection in Aerial Images Based on Cascaded Convolutional Neural Networks.

PubMed

Zhong, Jiandan; Lei, Tao; Yao, Guangle

2017-11-24

Vehicle detection in aerial images is an important and challenging task. Traditionally, many target detection models based on sliding-window fashion were developed and achieved acceptable performance, but these models are time-consuming in the detection phase. Recently, with the great success of convolutional neural networks (CNNs) in computer vision, many state-of-the-art detectors have been designed based on deep CNNs. However, these CNN-based detectors are inefficient when applied in aerial image data due to the fact that the existing CNN-based models struggle with small-size object detection and precise localization. To improve the detection accuracy without decreasing speed, we propose a CNN-based detection model combining two independent convolutional neural networks, where the first network is applied to generate a set of vehicle-like regions from multi-feature maps of different hierarchies and scales. Because the multi-feature maps combine the advantage of the deep and shallow convolutional layer, the first network performs well on locating the small targets in aerial image data. Then, the generated candidate regions are fed into the second network for feature extraction and decision making. Comprehensive experiments are conducted on the Vehicle Detection in Aerial Imagery (VEDAI) dataset and Munich vehicle dataset. The proposed cascaded detection model yields high performance, not only in detection accuracy but also in detection speed.

Robust Vehicle Detection in Aerial Images Based on Cascaded Convolutional Neural Networks

PubMed Central

Zhong, Jiandan; Lei, Tao; Yao, Guangle

2017-01-01

Vehicle detection in aerial images is an important and challenging task. Traditionally, many target detection models based on sliding-window fashion were developed and achieved acceptable performance, but these models are time-consuming in the detection phase. Recently, with the great success of convolutional neural networks (CNNs) in computer vision, many state-of-the-art detectors have been designed based on deep CNNs. However, these CNN-based detectors are inefficient when applied in aerial image data due to the fact that the existing CNN-based models struggle with small-size object detection and precise localization. To improve the detection accuracy without decreasing speed, we propose a CNN-based detection model combining two independent convolutional neural networks, where the first network is applied to generate a set of vehicle-like regions from multi-feature maps of different hierarchies and scales. Because the multi-feature maps combine the advantage of the deep and shallow convolutional layer, the first network performs well on locating the small targets in aerial image data. Then, the generated candidate regions are fed into the second network for feature extraction and decision making. Comprehensive experiments are conducted on the Vehicle Detection in Aerial Imagery (VEDAI) dataset and Munich vehicle dataset. The proposed cascaded detection model yields high performance, not only in detection accuracy but also in detection speed. PMID:29186756

Traffic sign recognition based on deep convolutional neural network

NASA Astrophysics Data System (ADS)

Yin, Shi-hao; Deng, Ji-cai; Zhang, Da-wei; Du, Jing-yuan

2017-11-01

Traffic sign recognition (TSR) is an important component of automated driving systems. It is a rather challenging task to design a high-performance classifier for the TSR system. In this paper, we propose a new method for TSR system based on deep convolutional neural network. In order to enhance the expression of the network, a novel structure (dubbed block-layer below) which combines network-in-network and residual connection is designed. Our network has 10 layers with parameters (block-layer seen as a single layer): the first seven are alternate convolutional layers and block-layers, and the remaining three are fully-connected layers. We train our TSR network on the German traffic sign recognition benchmark (GTSRB) dataset. To reduce overfitting, we perform data augmentation on the training images and employ a regularization method named "dropout". The activation function we employ in our network adopts scaled exponential linear units (SELUs), which can induce self-normalizing properties. To speed up the training, we use an efficient GPU to accelerate the convolutional operation. On the test dataset of GTSRB, we achieve the accuracy rate of 99.67%, exceeding the state-of-the-art results.

Improving deep convolutional neural networks with mixed maxout units

PubMed Central

Liu, Fu-xian; Li, Long-yue

2017-01-01

Motivated by insights from the maxout-units-based deep Convolutional Neural Network (CNN) that “non-maximal features are unable to deliver” and “feature mapping subspace pooling is insufficient,” we present a novel mixed variant of the recently introduced maxout unit called a mixout unit. Specifically, we do so by calculating the exponential probabilities of feature mappings gained by applying different convolutional transformations over the same input and then calculating the expected values according to their exponential probabilities. Moreover, we introduce the Bernoulli distribution to balance the maximum values with the expected values of the feature mappings subspace. Finally, we design a simple model to verify the pooling ability of mixout units and a Mixout-units-based Network-in-Network (NiN) model to analyze the feature learning ability of the mixout models. We argue that our proposed units improve the pooling ability and that mixout models can achieve better feature learning and classification performance. PMID:28727737

Convolutional neural network with transfer learning for rice type classification

NASA Astrophysics Data System (ADS)

Patel, Vaibhav Amit; Joshi, Manjunath V.

2018-04-01

Presently, rice type is identified manually by humans, which is time consuming and error prone. Therefore, there is a need to do this by machine which makes it faster with greater accuracy. This paper proposes a deep learning based method for classification of rice types. We propose two methods to classify the rice types. In the first method, we train a deep convolutional neural network (CNN) using the given segmented rice images. In the second method, we train a combination of a pretrained VGG16 network and the proposed method, while using transfer learning in which the weights of a pretrained network are used to achieve better accuracy. Our approach can also be used for classification of rice grain as broken or fine. We train a 5-class model for classifying rice types using 4000 training images and another 2- class model for the classification of broken and normal rice using 1600 training images. We observe that despite having distinct rice images, our architecture, pretrained on ImageNet data boosts classification accuracy significantly.

Accurate segmentation of lung fields on chest radiographs using deep convolutional networks

NASA Astrophysics Data System (ADS)

Arbabshirani, Mohammad R.; Dallal, Ahmed H.; Agarwal, Chirag; Patel, Aalpan; Moore, Gregory

2017-02-01

Accurate segmentation of lung fields on chest radiographs is the primary step for computer-aided detection of various conditions such as lung cancer and tuberculosis. The size, shape and texture of lung fields are key parameters for chest X-ray (CXR) based lung disease diagnosis in which the lung field segmentation is a significant primary step. Although many methods have been proposed for this problem, lung field segmentation remains as a challenge. In recent years, deep learning has shown state of the art performance in many visual tasks such as object detection, image classification and semantic image segmentation. In this study, we propose a deep convolutional neural network (CNN) framework for segmentation of lung fields. The algorithm was developed and tested on 167 clinical posterior-anterior (PA) CXR images collected retrospectively from picture archiving and communication system (PACS) of Geisinger Health System. The proposed multi-scale network is composed of five convolutional and two fully connected layers. The framework achieved IOU (intersection over union) of 0.96 on the testing dataset as compared to manual segmentation. The suggested framework outperforms state of the art registration-based segmentation by a significant margin. To our knowledge, this is the first deep learning based study of lung field segmentation on CXR images developed on a heterogeneous clinical dataset. The results suggest that convolutional neural networks could be employed reliably for lung field segmentation.

A deep learning method for lincRNA detection using auto-encoder algorithm.

PubMed

Yu, Ning; Yu, Zeng; Pan, Yi

2017-12-06

RNA sequencing technique (RNA-seq) enables scientists to develop novel data-driven methods for discovering more unidentified lincRNAs. Meantime, knowledge-based technologies are experiencing a potential revolution ignited by the new deep learning methods. By scanning the newly found data set from RNA-seq, scientists have found that: (1) the expression of lincRNAs appears to be regulated, that is, the relevance exists along the DNA sequences; (2) lincRNAs contain some conversed patterns/motifs tethered together by non-conserved regions. The two evidences give the reasoning for adopting knowledge-based deep learning methods in lincRNA detection. Similar to coding region transcription, non-coding regions are split at transcriptional sites. However, regulatory RNAs rather than message RNAs are generated. That is, the transcribed RNAs participate the biological process as regulatory units instead of generating proteins. Identifying these transcriptional regions from non-coding regions is the first step towards lincRNA recognition. The auto-encoder method achieves 100% and 92.4% prediction accuracy on transcription sites over the putative data sets. The experimental results also show the excellent performance of predictive deep neural network on the lincRNA data sets compared with support vector machine and traditional neural network. In addition, it is validated through the newly discovered lincRNA data set and one unreported transcription site is found by feeding the whole annotated sequences through the deep learning machine, which indicates that deep learning method has the extensive ability for lincRNA prediction. The transcriptional sequences of lincRNAs are collected from the annotated human DNA genome data. Subsequently, a two-layer deep neural network is developed for the lincRNA detection, which adopts the auto-encoder algorithm and utilizes different encoding schemes to obtain the best performance over intergenic DNA sequence data. Driven by those newly annotated lincRNA data, deep learning methods based on auto-encoder algorithm can exert their capability in knowledge learning in order to capture the useful features and the information correlation along DNA genome sequences for lincRNA detection. As our knowledge, this is the first application to adopt the deep learning techniques for identifying lincRNA transcription sequences.

Deep Learning with Convolutional Neural Networks Applied to Electromyography Data: A Resource for the Classification of Movements for Prosthetic Hands

PubMed Central

Atzori, Manfredo; Cognolato, Matteo; Müller, Henning

2016-01-01

Natural control methods based on surface electromyography (sEMG) and pattern recognition are promising for hand prosthetics. However, the control robustness offered by scientific research is still not sufficient for many real life applications, and commercial prostheses are capable of offering natural control for only a few movements. In recent years deep learning revolutionized several fields of machine learning, including computer vision and speech recognition. Our objective is to test its methods for natural control of robotic hands via sEMG using a large number of intact subjects and amputees. We tested convolutional networks for the classification of an average of 50 hand movements in 67 intact subjects and 11 transradial amputees. The simple architecture of the neural network allowed to make several tests in order to evaluate the effect of pre-processing, layer architecture, data augmentation and optimization. The classification results are compared with a set of classical classification methods applied on the same datasets. The classification accuracy obtained with convolutional neural networks using the proposed architecture is higher than the average results obtained with the classical classification methods, but lower than the results obtained with the best reference methods in our tests. The results show that convolutional neural networks with a very simple architecture can produce accurate results comparable to the average classical classification methods. They show that several factors (including pre-processing, the architecture of the net and the optimization parameters) can be fundamental for the analysis of sEMG data. Larger networks can achieve higher accuracy on computer vision and object recognition tasks. This fact suggests that it may be interesting to evaluate if larger networks can increase sEMG classification accuracy too. PMID:27656140

Deep Learning with Convolutional Neural Networks Applied to Electromyography Data: A Resource for the Classification of Movements for Prosthetic Hands.

PubMed

Atzori, Manfredo; Cognolato, Matteo; Müller, Henning

2016-01-01

Natural control methods based on surface electromyography (sEMG) and pattern recognition are promising for hand prosthetics. However, the control robustness offered by scientific research is still not sufficient for many real life applications, and commercial prostheses are capable of offering natural control for only a few movements. In recent years deep learning revolutionized several fields of machine learning, including computer vision and speech recognition. Our objective is to test its methods for natural control of robotic hands via sEMG using a large number of intact subjects and amputees. We tested convolutional networks for the classification of an average of 50 hand movements in 67 intact subjects and 11 transradial amputees. The simple architecture of the neural network allowed to make several tests in order to evaluate the effect of pre-processing, layer architecture, data augmentation and optimization. The classification results are compared with a set of classical classification methods applied on the same datasets. The classification accuracy obtained with convolutional neural networks using the proposed architecture is higher than the average results obtained with the classical classification methods, but lower than the results obtained with the best reference methods in our tests. The results show that convolutional neural networks with a very simple architecture can produce accurate results comparable to the average classical classification methods. They show that several factors (including pre-processing, the architecture of the net and the optimization parameters) can be fundamental for the analysis of sEMG data. Larger networks can achieve higher accuracy on computer vision and object recognition tasks. This fact suggests that it may be interesting to evaluate if larger networks can increase sEMG classification accuracy too.

Deep Direct Reinforcement Learning for Financial Signal Representation and Trading.

PubMed

Deng, Yue; Bao, Feng; Kong, Youyong; Ren, Zhiquan; Dai, Qionghai

2017-03-01

Can we train the computer to beat experienced traders for financial assert trading? In this paper, we try to address this challenge by introducing a recurrent deep neural network (NN) for real-time financial signal representation and trading. Our model is inspired by two biological-related learning concepts of deep learning (DL) and reinforcement learning (RL). In the framework, the DL part automatically senses the dynamic market condition for informative feature learning. Then, the RL module interacts with deep representations and makes trading decisions to accumulate the ultimate rewards in an unknown environment. The learning system is implemented in a complex NN that exhibits both the deep and recurrent structures. Hence, we propose a task-aware backpropagation through time method to cope with the gradient vanishing issue in deep training. The robustness of the neural system is verified on both the stock and the commodity future markets under broad testing conditions.

Deep learning on temporal-spectral data for anomaly detection

NASA Astrophysics Data System (ADS)

Ma, King; Leung, Henry; Jalilian, Ehsan; Huang, Daniel

2017-05-01

Detecting anomalies is important for continuous monitoring of sensor systems. One significant challenge is to use sensor data and autonomously detect changes that cause different conditions to occur. Using deep learning methods, we are able to monitor and detect changes as a result of some disturbance in the system. We utilize deep neural networks for sequence analysis of time series. We use a multi-step method for anomaly detection. We train the network to learn spectral and temporal features from the acoustic time series. We test our method using fiber-optic acoustic data from a pipeline.

Generalized Recurrent Neural Network accommodating Dynamic Causal Modeling for functional MRI analysis.

PubMed

Wang, Yuan; Wang, Yao; Lui, Yvonne W

2018-05-18

Dynamic Causal Modeling (DCM) is an advanced biophysical model which explicitly describes the entire process from experimental stimuli to functional magnetic resonance imaging (fMRI) signals via neural activity and cerebral hemodynamics. To conduct a DCM study, one needs to represent the experimental stimuli as a compact vector-valued function of time, which is hard in complex tasks such as book reading and natural movie watching. Deep learning provides the state-of-the-art signal representation solution, encoding complex signals into compact dense vectors while preserving the essence of the original signals. There is growing interest in using Recurrent Neural Networks (RNNs), a major family of deep learning techniques, in fMRI modeling. However, the generic RNNs used in existing studies work as black boxes, making the interpretation of results in a neuroscience context difficult and obscure. In this paper, we propose a new biophysically interpretable RNN built on DCM, DCM-RNN. We generalize the vanilla RNN and show that DCM can be cast faithfully as a special form of the generalized RNN. DCM-RNN uses back propagation for parameter estimation. We believe DCM-RNN is a promising tool for neuroscience. It can fit seamlessly into classical DCM studies. We demonstrate face validity of DCM-RNN in two principal applications of DCM: causal brain architecture hypotheses testing and effective connectivity estimation. We also demonstrate construct validity of DCM-RNN in an attention-visual experiment. Moreover, DCM-RNN enables end-to-end training of DCM and representation learning deep neural networks, extending DCM studies to complex tasks. Copyright © 2018 Elsevier Inc. All rights reserved.

Texture and art with deep neural networks.

PubMed

Gatys, Leon A; Ecker, Alexander S; Bethge, Matthias

2017-10-01

Although the study of biological vision and computer vision attempt to understand powerful visual information processing from different angles, they have a long history of informing each other. Recent advances in texture synthesis that were motivated by visual neuroscience have led to a substantial advance in image synthesis and manipulation in computer vision using convolutional neural networks (CNNs). Here, we review these recent advances and discuss how they can in turn inspire new research in visual perception and computational neuroscience. Copyright © 2017. Published by Elsevier Ltd.

Very Deep Convolutional Neural Networks for Morphologic Classification of Erythrocytes.

PubMed

Durant, Thomas J S; Olson, Eben M; Schulz, Wade L; Torres, Richard

2017-12-01

Morphologic profiling of the erythrocyte population is a widely used and clinically valuable diagnostic modality, but one that relies on a slow manual process associated with significant labor cost and limited reproducibility. Automated profiling of erythrocytes from digital images by capable machine learning approaches would augment the throughput and value of morphologic analysis. To this end, we sought to evaluate the performance of leading implementation strategies for convolutional neural networks (CNNs) when applied to classification of erythrocytes based on morphology. Erythrocytes were manually classified into 1 of 10 classes using a custom-developed Web application. Using recent literature to guide architectural considerations for neural network design, we implemented a "very deep" CNN, consisting of >150 layers, with dense shortcut connections. The final database comprised 3737 labeled cells. Ensemble model predictions on unseen data demonstrated a harmonic mean of recall and precision metrics of 92.70% and 89.39%, respectively. Of the 748 cells in the test set, 23 misclassification errors were made, with a correct classification frequency of 90.60%, represented as a harmonic mean across the 10 morphologic classes. These findings indicate that erythrocyte morphology profiles could be measured with a high degree of accuracy with "very deep" CNNs. Further, these data support future efforts to expand classes and optimize practical performance in a clinical environment as a prelude to full implementation as a clinical tool. © 2017 American Association for Clinical Chemistry.

The benefit of combining a deep neural network architecture with ideal ratio mask estimation in computational speech segregation to improve speech intelligibility.

PubMed

Bentsen, Thomas; May, Tobias; Kressner, Abigail A; Dau, Torsten

2018-01-01

Computational speech segregation attempts to automatically separate speech from noise. This is challenging in conditions with interfering talkers and low signal-to-noise ratios. Recent approaches have adopted deep neural networks and successfully demonstrated speech intelligibility improvements. A selection of components may be responsible for the success with these state-of-the-art approaches: the system architecture, a time frame concatenation technique and the learning objective. The aim of this study was to explore the roles and the relative contributions of these components by measuring speech intelligibility in normal-hearing listeners. A substantial improvement of 25.4 percentage points in speech intelligibility scores was found going from a subband-based architecture, in which a Gaussian Mixture Model-based classifier predicts the distributions of speech and noise for each frequency channel, to a state-of-the-art deep neural network-based architecture. Another improvement of 13.9 percentage points was obtained by changing the learning objective from the ideal binary mask, in which individual time-frequency units are labeled as either speech- or noise-dominated, to the ideal ratio mask, where the units are assigned a continuous value between zero and one. Therefore, both components play significant roles and by combining them, speech intelligibility improvements were obtained in a six-talker condition at a low signal-to-noise ratio.

Using deep convolutional neural networks to identify and classify tumor-associated stroma in diagnostic breast biopsies.

PubMed

Ehteshami Bejnordi, Babak; Mullooly, Maeve; Pfeiffer, Ruth M; Fan, Shaoqi; Vacek, Pamela M; Weaver, Donald L; Herschorn, Sally; Brinton, Louise A; van Ginneken, Bram; Karssemeijer, Nico; Beck, Andrew H; Gierach, Gretchen L; van der Laak, Jeroen A W M; Sherman, Mark E

2018-06-13

The breast stromal microenvironment is a pivotal factor in breast cancer development, growth and metastases. Although pathologists often detect morphologic changes in stroma by light microscopy, visual classification of such changes is subjective and non-quantitative, limiting its diagnostic utility. To gain insights into stromal changes associated with breast cancer, we applied automated machine learning techniques to digital images of 2387 hematoxylin and eosin stained tissue sections of benign and malignant image-guided breast biopsies performed to investigate mammographic abnormalities among 882 patients, ages 40-65 years, that were enrolled in the Breast Radiology Evaluation and Study of Tissues (BREAST) Stamp Project. Using deep convolutional neural networks, we trained an algorithm to discriminate between stroma surrounding invasive cancer and stroma from benign biopsies. In test sets (928 whole-slide images from 330 patients), this algorithm could distinguish biopsies diagnosed as invasive cancer from benign biopsies solely based on the stromal characteristics (area under the receiver operator characteristics curve = 0.962). Furthermore, without being trained specifically using ductal carcinoma in situ as an outcome, the algorithm detected tumor-associated stroma in greater amounts and at larger distances from grade 3 versus grade 1 ductal carcinoma in situ. Collectively, these results suggest that algorithms based on deep convolutional neural networks that evaluate only stroma may prove useful to classify breast biopsies and aid in understanding and evaluating the biology of breast lesions.

Deep neural network using color and synthesized three-dimensional shape for face recognition

NASA Astrophysics Data System (ADS)

Rhee, Seon-Min; Yoo, ByungIn; Han, Jae-Joon; Hwang, Wonjun

2017-03-01

We present an approach for face recognition using synthesized three-dimensional (3-D) shape information together with two-dimensional (2-D) color in a deep convolutional neural network (DCNN). As 3-D facial shape is hardly affected by the extrinsic 2-D texture changes caused by illumination, make-up, and occlusions, it could provide more reliable complementary features in harmony with the 2-D color feature in face recognition. Unlike other approaches that use 3-D shape information with the help of an additional depth sensor, our approach generates a personalized 3-D face model by using only face landmarks in the 2-D input image. Using the personalized 3-D face model, we generate a frontalized 2-D color facial image as well as 3-D facial images (e.g., a depth image and a normal image). In our DCNN, we first feed 2-D and 3-D facial images into independent convolutional layers, where the low-level kernels are successfully learned according to their own characteristics. Then, we merge them and feed into higher-level layers under a single deep neural network. Our proposed approach is evaluated with labeled faces in the wild dataset and the results show that the error rate of the verification rate at false acceptance rate 1% is improved by up to 32.1% compared with the baseline where only a 2-D color image is used.

An automatic microseismic or acoustic emission arrival identification scheme with deep recurrent neural networks

NASA Astrophysics Data System (ADS)

Zheng, Jing; Lu, Jiren; Peng, Suping; Jiang, Tianqi

2018-02-01

The conventional arrival pick-up algorithms cannot avoid the manual modification of the parameters for the simultaneous identification of multiple events under different signal-to-noise ratios (SNRs). Therefore, in order to automatically obtain the arrivals of multiple events with high precision under different SNRs, in this study an algorithm was proposed which had the ability to pick up the arrival of microseismic or acoustic emission events based on deep recurrent neural networks. The arrival identification was performed using two important steps, which included a training phase and a testing phase. The training process was mathematically modelled by deep recurrent neural networks using Long Short-Term Memory architecture. During the testing phase, the learned weights were utilized to identify the arrivals through the microseismic/acoustic emission data sets. The data sets were obtained by rock physics experiments of the acoustic emission. In order to obtain the data sets under different SNRs, this study added random noise to the raw experiments' data sets. The results showed that the outcome of the proposed method was able to attain an above 80 per cent hit-rate at SNR 0 dB, and an approximately 70 per cent hit-rate at SNR -5 dB, with an absolute error in 10 sampling points. These results indicated that the proposed method had high selection precision and robustness.

Complex Rotation Quantum Dynamic Neural Networks (CRQDNN) using Complex Quantum Neuron (CQN): Applications to time series prediction.

PubMed

Cui, Yiqian; Shi, Junyou; Wang, Zili

2015-11-01

Quantum Neural Networks (QNN) models have attracted great attention since it innovates a new neural computing manner based on quantum entanglement. However, the existing QNN models are mainly based on the real quantum operations, and the potential of quantum entanglement is not fully exploited. In this paper, we proposes a novel quantum neuron model called Complex Quantum Neuron (CQN) that realizes a deep quantum entanglement. Also, a novel hybrid networks model Complex Rotation Quantum Dynamic Neural Networks (CRQDNN) is proposed based on Complex Quantum Neuron (CQN). CRQDNN is a three layer model with both CQN and classical neurons. An infinite impulse response (IIR) filter is embedded in the Networks model to enable the memory function to process time series inputs. The Levenberg-Marquardt (LM) algorithm is used for fast parameter learning. The networks model is developed to conduct time series predictions. Two application studies are done in this paper, including the chaotic time series prediction and electronic remaining useful life (RUL) prediction. Copyright © 2015 Elsevier Ltd. All rights reserved.

Staged Inference using Conditional Deep Learning for energy efficient real-time smart diagnosis.

PubMed

Parsa, Maryam; Panda, Priyadarshini; Sen, Shreyas; Roy, Kaushik

2017-07-01

Recent progress in biosensor technology and wearable devices has created a formidable opportunity for remote healthcare monitoring systems as well as real-time diagnosis and disease prevention. The use of data mining techniques is indispensable for analysis of the large pool of data generated by the wearable devices. Deep learning is among the promising methods for analyzing such data for healthcare applications and disease diagnosis. However, the conventional deep neural networks are computationally intensive and it is impractical to use them in real-time diagnosis with low-powered on-body devices. We propose Staged Inference using Conditional Deep Learning (SICDL), as an energy efficient approach for creating healthcare monitoring systems. For smart diagnostics, we observe that all diagnoses are not equally challenging. The proposed approach thus decomposes the diagnoses into preliminary analysis (such as healthy vs unhealthy) and detailed analysis (such as identifying the specific type of cardio disease). The preliminary diagnosis is conducted real-time with a low complexity neural network realized on the resource-constrained on-body device. The detailed diagnosis requires a larger network that is implemented remotely in cloud and is conditionally activated only for detailed diagnosis (unhealthy individuals). We evaluated the proposed approach using available physiological sensor data from Physionet databases, and achieved 38% energy reduction in comparison to the conventional deep learning approach.

Deep Learning for Computer Vision: A Brief Review

PubMed Central

Doulamis, Nikolaos; Doulamis, Anastasios; Protopapadakis, Eftychios

2018-01-01

Over the last years deep learning methods have been shown to outperform previous state-of-the-art machine learning techniques in several fields, with computer vision being one of the most prominent cases. This review paper provides a brief overview of some of the most significant deep learning schemes used in computer vision problems, that is, Convolutional Neural Networks, Deep Boltzmann Machines and Deep Belief Networks, and Stacked Denoising Autoencoders. A brief account of their history, structure, advantages, and limitations is given, followed by a description of their applications in various computer vision tasks, such as object detection, face recognition, action and activity recognition, and human pose estimation. Finally, a brief overview is given of future directions in designing deep learning schemes for computer vision problems and the challenges involved therein. PMID:29487619

Detection of Interictal Discharges With Convolutional Neural Networks Using Discrete Ordered Multichannel Intracranial EEG.

PubMed

Antoniades, Andreas; Spyrou, Loukianos; Martin-Lopez, David; Valentin, Antonio; Alarcon, Gonzalo; Sanei, Saeid; Cheong Took, Clive

2017-12-01

Detection algorithms for electroencephalography (EEG) data, especially in the field of interictal epileptiform discharge (IED) detection, have traditionally employed handcrafted features, which utilized specific characteristics of neural responses. Although these algorithms achieve high accuracy, mere detection of an IED holds little clinical significance. In this paper, we consider deep learning for epileptic subjects to accommodate automatic feature generation from intracranial EEG data, while also providing clinical insight. Convolutional neural networks are trained in a subject independent fashion to demonstrate how meaningful features are automatically learned in a hierarchical process. We illustrate how the convolved filters in the deepest layers provide insight toward the different types of IEDs within the group, as confirmed by our expert clinicians. The morphology of the IEDs found in filters can help evaluate the treatment of a patient. To improve the learning of the deep model, moderately different score classes are utilized as opposed to binary IED and non-IED labels. The resulting model achieves state-of-the-art classification performance and is also invariant to time differences between the IEDs. This paper suggests that deep learning is suitable for automatic feature generation from intracranial EEG data, while also providing insight into the data.

Diagnosis and prediction of periodontally compromised teeth using a deep learning-based convolutional neural network algorithm.

PubMed

Lee, Jae-Hong; Kim, Do-Hyung; Jeong, Seong-Nyum; Choi, Seong-Ho

2018-04-01

The aim of the current study was to develop a computer-assisted detection system based on a deep convolutional neural network (CNN) algorithm and to evaluate the potential usefulness and accuracy of this system for the diagnosis and prediction of periodontally compromised teeth (PCT). Combining pretrained deep CNN architecture and a self-trained network, periapical radiographic images were used to determine the optimal CNN algorithm and weights. The diagnostic and predictive accuracy, sensitivity, specificity, positive predictive value, negative predictive value, receiver operating characteristic (ROC) curve, area under the ROC curve, confusion matrix, and 95% confidence intervals (CIs) were calculated using our deep CNN algorithm, based on a Keras framework in Python. The periapical radiographic dataset was split into training (n=1,044), validation (n=348), and test (n=348) datasets. With the deep learning algorithm, the diagnostic accuracy for PCT was 81.0% for premolars and 76.7% for molars. Using 64 premolars and 64 molars that were clinically diagnosed as severe PCT, the accuracy of predicting extraction was 82.8% (95% CI, 70.1%-91.2%) for premolars and 73.4% (95% CI, 59.9%-84.0%) for molars. We demonstrated that the deep CNN algorithm was useful for assessing the diagnosis and predictability of PCT. Therefore, with further optimization of the PCT dataset and improvements in the algorithm, a computer-aided detection system can be expected to become an effective and efficient method of diagnosing and predicting PCT.

First Time Rapid and Accurate Detection of Massive Number of Metal Absorption Lines in the Early Universe Using Deep Neural Network

NASA Astrophysics Data System (ADS)

Zhao, Yinan; Ge, Jian; Yuan, Xiaoyong; Li, Xiaolin; Zhao, Tiffany; Wang, Cindy

2018-01-01

Metal absorption line systems in the distant quasar spectra have been used as one of the most powerful tools to probe gas content in the early Universe. The MgII λλ 2796, 2803 doublet is one of the most popular metal absorption lines and has been used to trace gas and global star formation at redshifts between ~0.5 to 2.5. In the past, machine learning algorithms have been used to detect absorption lines systems in the large sky survey, such as Principle Component Analysis, Gaussian Process and decision tree, but the overall detection process is not only complicated, but also time consuming. It usually takes a few months to go through the entire quasar spectral dataset from each of the Sloan Digital Sky Survey (SDSS) data release. In this work, we applied the deep neural network, or “ deep learning” algorithms, in the most recently SDSS DR14 quasar spectra and were able to randomly search 20000 quasar spectra and detect 2887 strong Mg II absorption features in just 9 seconds. Our detection algorithms were verified with previously released DR12 and DR7 data and published Mg II catalog and the detection accuracy is 90%. This is the first time that deep neural network has demonstrated its promising power in both speed and accuracy in replacing tedious, repetitive human work in searching for narrow absorption patterns in a big dataset. We will present our detection algorithms and also statistical results of the newly detected Mg II absorption lines.

Evolutionary pruning of transfer learned deep convolutional neural network for breast cancer diagnosis in digital breast tomosynthesis.

PubMed

Samala, Ravi K; Chan, Heang-Ping; Hadjiiski, Lubomir M; Helvie, Mark A; Richter, Caleb; Cha, Kenny

2018-05-01

Deep learning models are highly parameterized, resulting in difficulty in inference and transfer learning for image recognition tasks. In this work, we propose a layered pathway evolution method to compress a deep convolutional neural network (DCNN) for classification of masses in digital breast tomosynthesis (DBT). The objective is to prune the number of tunable parameters while preserving the classification accuracy. In the first stage transfer learning, 19 632 augmented regions-of-interest (ROIs) from 2454 mass lesions on mammograms were used to train a pre-trained DCNN on ImageNet. In the second stage transfer learning, the DCNN was used as a feature extractor followed by feature selection and random forest classification. The pathway evolution was performed using genetic algorithm in an iterative approach with tournament selection driven by count-preserving crossover and mutation. The second stage was trained with 9120 DBT ROIs from 228 mass lesions using leave-one-case-out cross-validation. The DCNN was reduced by 87% in the number of neurons, 34% in the number of parameters, and 95% in the number of multiply-and-add operations required in the convolutional layers. The test AUC on 89 mass lesions from 94 independent DBT cases before and after pruning were 0.88 and 0.90, respectively, and the difference was not statistically significant (p  >  0.05). The proposed DCNN compression approach can reduce the number of required operations by 95% while maintaining the classification performance. The approach can be extended to other deep neural networks and imaging tasks where transfer learning is appropriate.

Evolutionary pruning of transfer learned deep convolutional neural network for breast cancer diagnosis in digital breast tomosynthesis

NASA Astrophysics Data System (ADS)

Samala, Ravi K.; Chan, Heang-Ping; Hadjiiski, Lubomir M.; Helvie, Mark A.; Richter, Caleb; Cha, Kenny

2018-05-01

Deep learning models are highly parameterized, resulting in difficulty in inference and transfer learning for image recognition tasks. In this work, we propose a layered pathway evolution method to compress a deep convolutional neural network (DCNN) for classification of masses in digital breast tomosynthesis (DBT). The objective is to prune the number of tunable parameters while preserving the classification accuracy. In the first stage transfer learning, 19 632 augmented regions-of-interest (ROIs) from 2454 mass lesions on mammograms were used to train a pre-trained DCNN on ImageNet. In the second stage transfer learning, the DCNN was used as a feature extractor followed by feature selection and random forest classification. The pathway evolution was performed using genetic algorithm in an iterative approach with tournament selection driven by count-preserving crossover and mutation. The second stage was trained with 9120 DBT ROIs from 228 mass lesions using leave-one-case-out cross-validation. The DCNN was reduced by 87% in the number of neurons, 34% in the number of parameters, and 95% in the number of multiply-and-add operations required in the convolutional layers. The test AUC on 89 mass lesions from 94 independent DBT cases before and after pruning were 0.88 and 0.90, respectively, and the difference was not statistically significant (p  >  0.05). The proposed DCNN compression approach can reduce the number of required operations by 95% while maintaining the classification performance. The approach can be extended to other deep neural networks and imaging tasks where transfer learning is appropriate.

Classification of brain MRI with big data and deep 3D convolutional neural networks

NASA Astrophysics Data System (ADS)

Wegmayr, Viktor; Aitharaju, Sai; Buhmann, Joachim

2018-02-01

Our ever-aging society faces the growing problem of neurodegenerative diseases, in particular dementia. Magnetic Resonance Imaging provides a unique tool for non-invasive investigation of these brain diseases. However, it is extremely difficult for neurologists to identify complex disease patterns from large amounts of three-dimensional images. In contrast, machine learning excels at automatic pattern recognition from large amounts of data. In particular, deep learning has achieved impressive results in image classification. Unfortunately, its application to medical image classification remains difficult. We consider two reasons for this difficulty: First, volumetric medical image data is considerably scarcer than natural images. Second, the complexity of 3D medical images is much higher compared to common 2D images. To address the problem of small data set size, we assemble the largest dataset ever used for training a deep 3D convolutional neural network to classify brain images as healthy (HC), mild cognitive impairment (MCI) or Alzheimers disease (AD). We use more than 20.000 images from subjects of these three classes, which is almost 9x the size of the previously largest data set. The problem of high dimensionality is addressed by using a deep 3D convolutional neural network, which is state-of-the-art in large-scale image classification. We exploit its ability to process the images directly, only with standard preprocessing, but without the need for elaborate feature engineering. Compared to other work, our workflow is considerably simpler, which increases clinical applicability. Accuracy is measured on the ADNI+AIBL data sets, and the independent CADDementia benchmark.

Deep linear autoencoder and patch clustering-based unified one-dimensional coding of image and video

NASA Astrophysics Data System (ADS)

Li, Honggui

2017-09-01

This paper proposes a unified one-dimensional (1-D) coding framework of image and video, which depends on deep learning neural network and image patch clustering. First, an improved K-means clustering algorithm for image patches is employed to obtain the compact inputs of deep artificial neural network. Second, for the purpose of best reconstructing original image patches, deep linear autoencoder (DLA), a linear version of the classical deep nonlinear autoencoder, is introduced to achieve the 1-D representation of image blocks. Under the circumstances of 1-D representation, DLA is capable of attaining zero reconstruction error, which is impossible for the classical nonlinear dimensionality reduction methods. Third, a unified 1-D coding infrastructure for image, intraframe, interframe, multiview video, three-dimensional (3-D) video, and multiview 3-D video is built by incorporating different categories of videos into the inputs of patch clustering algorithm. Finally, it is shown in the results of simulation experiments that the proposed methods can simultaneously gain higher compression ratio and peak signal-to-noise ratio than those of the state-of-the-art methods in the situation of low bitrate transmission.

Cell segmentation in histopathological images with deep learning algorithms by utilizing spatial relationships.

PubMed

Hatipoglu, Nuh; Bilgin, Gokhan

2017-10-01

In many computerized methods for cell detection, segmentation, and classification in digital histopathology that have recently emerged, the task of cell segmentation remains a chief problem for image processing in designing computer-aided diagnosis (CAD) systems. In research and diagnostic studies on cancer, pathologists can use CAD systems as second readers to analyze high-resolution histopathological images. Since cell detection and segmentation are critical for cancer grade assessments, cellular and extracellular structures should primarily be extracted from histopathological images. In response, we sought to identify a useful cell segmentation approach with histopathological images that uses not only prominent deep learning algorithms (i.e., convolutional neural networks, stacked autoencoders, and deep belief networks), but also spatial relationships, information of which is critical for achieving better cell segmentation results. To that end, we collected cellular and extracellular samples from histopathological images by windowing in small patches with various sizes. In experiments, the segmentation accuracies of the methods used improved as the window sizes increased due to the addition of local spatial and contextual information. Once we compared the effects of training sample size and influence of window size, results revealed that the deep learning algorithms, especially convolutional neural networks and partly stacked autoencoders, performed better than conventional methods in cell segmentation.

Deep Artificial Neural Networks and Neuromorphic Chips for Big Data Analysis: Pharmaceutical and Bioinformatics Applications.

PubMed

Pastur-Romay, Lucas Antón; Cedrón, Francisco; Pazos, Alejandro; Porto-Pazos, Ana Belén

2016-08-11

Over the past decade, Deep Artificial Neural Networks (DNNs) have become the state-of-the-art algorithms in Machine Learning (ML), speech recognition, computer vision, natural language processing and many other tasks. This was made possible by the advancement in Big Data, Deep Learning (DL) and drastically increased chip processing abilities, especially general-purpose graphical processing units (GPGPUs). All this has created a growing interest in making the most of the potential offered by DNNs in almost every field. An overview of the main architectures of DNNs, and their usefulness in Pharmacology and Bioinformatics are presented in this work. The featured applications are: drug design, virtual screening (VS), Quantitative Structure-Activity Relationship (QSAR) research, protein structure prediction and genomics (and other omics) data mining. The future need of neuromorphic hardware for DNNs is also discussed, and the two most advanced chips are reviewed: IBM TrueNorth and SpiNNaker. In addition, this review points out the importance of considering not only neurons, as DNNs and neuromorphic chips should also include glial cells, given the proven importance of astrocytes, a type of glial cell which contributes to information processing in the brain. The Deep Artificial Neuron-Astrocyte Networks (DANAN) could overcome the difficulties in architecture design, learning process and scalability of the current ML methods.

Deep Artificial Neural Networks and Neuromorphic Chips for Big Data Analysis: Pharmaceutical and Bioinformatics Applications

PubMed Central

Pastur-Romay, Lucas Antón; Cedrón, Francisco; Pazos, Alejandro; Porto-Pazos, Ana Belén

2016-01-01

Over the past decade, Deep Artificial Neural Networks (DNNs) have become the state-of-the-art algorithms in Machine Learning (ML), speech recognition, computer vision, natural language processing and many other tasks. This was made possible by the advancement in Big Data, Deep Learning (DL) and drastically increased chip processing abilities, especially general-purpose graphical processing units (GPGPUs). All this has created a growing interest in making the most of the potential offered by DNNs in almost every field. An overview of the main architectures of DNNs, and their usefulness in Pharmacology and Bioinformatics are presented in this work. The featured applications are: drug design, virtual screening (VS), Quantitative Structure–Activity Relationship (QSAR) research, protein structure prediction and genomics (and other omics) data mining. The future need of neuromorphic hardware for DNNs is also discussed, and the two most advanced chips are reviewed: IBM TrueNorth and SpiNNaker. In addition, this review points out the importance of considering not only neurons, as DNNs and neuromorphic chips should also include glial cells, given the proven importance of astrocytes, a type of glial cell which contributes to information processing in the brain. The Deep Artificial Neuron–Astrocyte Networks (DANAN) could overcome the difficulties in architecture design, learning process and scalability of the current ML methods. PMID:27529225

Volumetric multimodality neural network for brain tumor segmentation

NASA Astrophysics Data System (ADS)

Silvana Castillo, Laura; Alexandra Daza, Laura; Carlos Rivera, Luis; Arbeláez, Pablo

2017-11-01

Brain lesion segmentation is one of the hardest tasks to be solved in computer vision with an emphasis on the medical field. We present a convolutional neural network that produces a semantic segmentation of brain tumors, capable of processing volumetric data along with information from multiple MRI modalities at the same time. This results in the ability to learn from small training datasets and highly imbalanced data. Our method is based on DeepMedic, the state of the art in brain lesion segmentation. We develop a new architecture with more convolutional layers, organized in three parallel pathways with different input resolution, and additional fully connected layers. We tested our method over the 2015 BraTS Challenge dataset, reaching an average dice coefficient of 84%, while the standard DeepMedic implementation reached 74%.

PSNet: prostate segmentation on MRI based on a convolutional neural network.

PubMed

Tian, Zhiqiang; Liu, Lizhi; Zhang, Zhenfeng; Fei, Baowei

2018-04-01

Automatic segmentation of the prostate on magnetic resonance images (MRI) has many applications in prostate cancer diagnosis and therapy. We proposed a deep fully convolutional neural network (CNN) to segment the prostate automatically. Our deep CNN model is trained end-to-end in a single learning stage, which uses prostate MRI and the corresponding ground truths as inputs. The learned CNN model can be used to make an inference for pixel-wise segmentation. Experiments were performed on three data sets, which contain prostate MRI of 140 patients. The proposed CNN model of prostate segmentation (PSNet) obtained a mean Dice similarity coefficient of [Formula: see text] as compared to the manually labeled ground truth. Experimental results show that the proposed model could yield satisfactory segmentation of the prostate on MRI.

Computational optical tomography using 3-D deep convolutional neural networks

NASA Astrophysics Data System (ADS)

Nguyen, Thanh; Bui, Vy; Nehmetallah, George

2018-04-01

Deep convolutional neural networks (DCNNs) offer a promising performance for many image processing areas, such as super-resolution, deconvolution, image classification, denoising, and segmentation, with outstanding results. Here, we develop for the first time, to our knowledge, a method to perform 3-D computational optical tomography using 3-D DCNN. A simulated 3-D phantom dataset was first constructed and converted to a dataset of phase objects imaged on a spatial light modulator. For each phase image in the dataset, the corresponding diffracted intensity image was experimentally recorded on a CCD. We then experimentally demonstrate the ability of the developed 3-D DCNN algorithm to solve the inverse problem by reconstructing the 3-D index of refraction distributions of test phantoms from the dataset from their corresponding diffraction patterns.

Korean letter handwritten recognition using deep convolutional neural network on android platform

NASA Astrophysics Data System (ADS)

Purnamawati, S.; Rachmawati, D.; Lumanauw, G.; Rahmat, R. F.; Taqyuddin, R.

2018-03-01

Currently, popularity of Korean culture attracts many people to learn everything about Korea, particularly its language. To acquire Korean Language, every single learner needs to be able to understand Korean non-Latin character. A digital approach needs to be carried out in order to make Korean learning process easier. This study is done by using Deep Convolutional Neural Network (DCNN). DCNN performs the recognition process on the image based on the model that has been trained such as Inception-v3 Model. Subsequently, re-training process using transfer learning technique with the trained and re-trained value of model is carried though in order to develop a new model with a better performance without any specific systemic errors. The testing accuracy of this research results in 86,9%.

Distinct contributions of functional and deep neural network features to representational similarity of scenes in human brain and behavior

PubMed Central

Greene, Michelle R; Baldassano, Christopher; Fei-Fei, Li; Beck, Diane M; Baker, Chris I

2018-01-01

Inherent correlations between visual and semantic features in real-world scenes make it difficult to determine how different scene properties contribute to neural representations. Here, we assessed the contributions of multiple properties to scene representation by partitioning the variance explained in human behavioral and brain measurements by three feature models whose inter-correlations were minimized a priori through stimulus preselection. Behavioral assessments of scene similarity reflected unique contributions from a functional feature model indicating potential actions in scenes as well as high-level visual features from a deep neural network (DNN). In contrast, similarity of cortical responses in scene-selective areas was uniquely explained by mid- and high-level DNN features only, while an object label model did not contribute uniquely to either domain. The striking dissociation between functional and DNN features in their contribution to behavioral and brain representations of scenes indicates that scene-selective cortex represents only a subset of behaviorally relevant scene information. PMID:29513219

Distinct contributions of functional and deep neural network features to representational similarity of scenes in human brain and behavior.

PubMed

Groen, Iris Ia; Greene, Michelle R; Baldassano, Christopher; Fei-Fei, Li; Beck, Diane M; Baker, Chris I

2018-03-07

Inherent correlations between visual and semantic features in real-world scenes make it difficult to determine how different scene properties contribute to neural representations. Here, we assessed the contributions of multiple properties to scene representation by partitioning the variance explained in human behavioral and brain measurements by three feature models whose inter-correlations were minimized a priori through stimulus preselection. Behavioral assessments of scene similarity reflected unique contributions from a functional feature model indicating potential actions in scenes as well as high-level visual features from a deep neural network (DNN). In contrast, similarity of cortical responses in scene-selective areas was uniquely explained by mid- and high-level DNN features only, while an object label model did not contribute uniquely to either domain. The striking dissociation between functional and DNN features in their contribution to behavioral and brain representations of scenes indicates that scene-selective cortex represents only a subset of behaviorally relevant scene information.

AUC-Maximized Deep Convolutional Neural Fields for Protein Sequence Labeling.

PubMed

Wang, Sheng; Sun, Siqi; Xu, Jinbo

2016-09-01

Deep Convolutional Neural Networks (DCNN) has shown excellent performance in a variety of machine learning tasks. This paper presents Deep Convolutional Neural Fields (DeepCNF), an integration of DCNN with Conditional Random Field (CRF), for sequence labeling with an imbalanced label distribution. The widely-used training methods, such as maximum-likelihood and maximum labelwise accuracy, do not work well on imbalanced data. To handle this, we present a new training algorithm called maximum-AUC for DeepCNF. That is, we train DeepCNF by directly maximizing the empirical Area Under the ROC Curve (AUC), which is an unbiased measurement for imbalanced data. To fulfill this, we formulate AUC in a pairwise ranking framework, approximate it by a polynomial function and then apply a gradient-based procedure to optimize it. Our experimental results confirm that maximum-AUC greatly outperforms the other two training methods on 8-state secondary structure prediction and disorder prediction since their label distributions are highly imbalanced and also has similar performance as the other two training methods on solvent accessibility prediction, which has three equally-distributed labels. Furthermore, our experimental results show that our AUC-trained DeepCNF models greatly outperform existing popular predictors of these three tasks. The data and software related to this paper are available at https://github.com/realbigws/DeepCNF_AUC.

AUC-Maximized Deep Convolutional Neural Fields for Protein Sequence Labeling

PubMed Central

Wang, Sheng; Sun, Siqi

2017-01-01

Deep Convolutional Neural Networks (DCNN) has shown excellent performance in a variety of machine learning tasks. This paper presents Deep Convolutional Neural Fields (DeepCNF), an integration of DCNN with Conditional Random Field (CRF), for sequence labeling with an imbalanced label distribution. The widely-used training methods, such as maximum-likelihood and maximum labelwise accuracy, do not work well on imbalanced data. To handle this, we present a new training algorithm called maximum-AUC for DeepCNF. That is, we train DeepCNF by directly maximizing the empirical Area Under the ROC Curve (AUC), which is an unbiased measurement for imbalanced data. To fulfill this, we formulate AUC in a pairwise ranking framework, approximate it by a polynomial function and then apply a gradient-based procedure to optimize it. Our experimental results confirm that maximum-AUC greatly outperforms the other two training methods on 8-state secondary structure prediction and disorder prediction since their label distributions are highly imbalanced and also has similar performance as the other two training methods on solvent accessibility prediction, which has three equally-distributed labels. Furthermore, our experimental results show that our AUC-trained DeepCNF models greatly outperform existing popular predictors of these three tasks. The data and software related to this paper are available at https://github.com/realbigws/DeepCNF_AUC. PMID:28884168

Deep learning and face recognition: the state of the art

NASA Astrophysics Data System (ADS)

Balaban, Stephen

2015-05-01

Deep Neural Networks (DNNs) have established themselves as a dominant technique in machine learning. DNNs have been top performers on a wide variety of tasks including image classification, speech recognition, and face recognition.1-3 Convolutional neural networks (CNNs) have been used in nearly all of the top performing methods on the Labeled Faces in the Wild (LFW) dataset.3-6 In this talk and accompanying paper, I attempt to provide a review and summary of the deep learning techniques used in the state-of-the-art. In addition, I highlight the need for both larger and more challenging public datasets to benchmark these systems. Despite the ability of DNNs and autoencoders to perform unsupervised feature learning, modern facial recognition pipelines still require domain specific engineering in the form of re-alignment. For example, in Facebook's recent DeepFace paper, a 3D "frontalization" step lies at the beginning of the pipeline. This step creates a 3D face model for the incoming image and then uses a series of affine transformations of the fiducial points to "frontalize" the image. This step enables the DeepFace system to use a neural network architecture with locally connected layers without weight sharing as opposed to standard convolutional layers.6 Deep learning techniques combined with large datasets have allowed research groups to surpass human level performance on the LFW dataset.3, 5 The high accuracy (99.63% for FaceNet at the time of publishing) and utilization of outside data (hundreds of millions of images in the case of Google's FaceNet) suggest that current face verification benchmarks such as LFW may not be challenging enough, nor provide enough data, for current techniques.3, 5 There exist a variety of organizations with mobile photo sharing applications that would be capable of releasing a very large scale and highly diverse dataset of facial images captured on mobile devices. Such an "ImageNet for Face Recognition" would likely receive a warm welcome from researchers and practitioners alike.

Chinese Sentence Classification Based on Convolutional Neural Network

NASA Astrophysics Data System (ADS)

Gu, Chengwei; Wu, Ming; Zhang, Chuang

2017-10-01

Sentence classification is one of the significant issues in Natural Language Processing (NLP). Feature extraction is often regarded as the key point for natural language processing. Traditional ways based on machine learning can not take high level features into consideration, such as Naive Bayesian Model. The neural network for sentence classification can make use of contextual information to achieve greater results in sentence classification tasks. In this paper, we focus on classifying Chinese sentences. And the most important is that we post a novel architecture of Convolutional Neural Network (CNN) to apply on Chinese sentence classification. In particular, most of the previous methods often use softmax classifier for prediction, we embed a linear support vector machine to substitute softmax in the deep neural network model, minimizing a margin-based loss to get a better result. And we use tanh as an activation function, instead of ReLU. The CNN model improve the result of Chinese sentence classification tasks. Experimental results on the Chinese news title database validate the effectiveness of our model.

Mastering the game of Go without human knowledge.

PubMed

Silver, David; Schrittwieser, Julian; Simonyan, Karen; Antonoglou, Ioannis; Huang, Aja; Guez, Arthur; Hubert, Thomas; Baker, Lucas; Lai, Matthew; Bolton, Adrian; Chen, Yutian; Lillicrap, Timothy; Hui, Fan; Sifre, Laurent; van den Driessche, George; Graepel, Thore; Hassabis, Demis

2017-10-18

A long-standing goal of artificial intelligence is an algorithm that learns, tabula rasa, superhuman proficiency in challenging domains. Recently, AlphaGo became the first program to defeat a world champion in the game of Go. The tree search in AlphaGo evaluated positions and selected moves using deep neural networks. These neural networks were trained by supervised learning from human expert moves, and by reinforcement learning from self-play. Here we introduce an algorithm based solely on reinforcement learning, without human data, guidance or domain knowledge beyond game rules. AlphaGo becomes its own teacher: a neural network is trained to predict AlphaGo's own move selections and also the winner of AlphaGo's games. This neural network improves the strength of the tree search, resulting in higher quality move selection and stronger self-play in the next iteration. Starting tabula rasa, our new program AlphaGo Zero achieved superhuman performance, winning 100-0 against the previously published, champion-defeating AlphaGo.

Mastering the game of Go without human knowledge

NASA Astrophysics Data System (ADS)

Silver, David; Schrittwieser, Julian; Simonyan, Karen; Antonoglou, Ioannis; Huang, Aja; Guez, Arthur; Hubert, Thomas; Baker, Lucas; Lai, Matthew; Bolton, Adrian; Chen, Yutian; Lillicrap, Timothy; Hui, Fan; Sifre, Laurent; van den Driessche, George; Graepel, Thore; Hassabis, Demis

2017-10-01

A long-standing goal of artificial intelligence is an algorithm that learns, tabula rasa, superhuman proficiency in challenging domains. Recently, AlphaGo became the first program to defeat a world champion in the game of Go. The tree search in AlphaGo evaluated positions and selected moves using deep neural networks. These neural networks were trained by supervised learning from human expert moves, and by reinforcement learning from self-play. Here we introduce an algorithm based solely on reinforcement learning, without human data, guidance or domain knowledge beyond game rules. AlphaGo becomes its own teacher: a neural network is trained to predict AlphaGo’s own move selections and also the winner of AlphaGo’s games. This neural network improves the strength of the tree search, resulting in higher quality move selection and stronger self-play in the next iteration. Starting tabula rasa, our new program AlphaGo Zero achieved superhuman performance, winning 100-0 against the previously published, champion-defeating AlphaGo.

A sparse autoencoder-based deep neural network for protein solvent accessibility and contact number prediction.

PubMed

Deng, Lei; Fan, Chao; Zeng, Zhiwen

2017-12-28

Direct prediction of the three-dimensional (3D) structures of proteins from one-dimensional (1D) sequences is a challenging problem. Significant structural characteristics such as solvent accessibility and contact number are essential for deriving restrains in modeling protein folding and protein 3D structure. Thus, accurately predicting these features is a critical step for 3D protein structure building. In this study, we present DeepSacon, a computational method that can effectively predict protein solvent accessibility and contact number by using a deep neural network, which is built based on stacked autoencoder and a dropout method. The results demonstrate that our proposed DeepSacon achieves a significant improvement in the prediction quality compared with the state-of-the-art methods. We obtain 0.70 three-state accuracy for solvent accessibility, 0.33 15-state accuracy and 0.74 Pearson Correlation Coefficient (PCC) for the contact number on the 5729 monomeric soluble globular protein dataset. We also evaluate the performance on the CASP11 benchmark dataset, DeepSacon achieves 0.68 three-state accuracy and 0.69 PCC for solvent accessibility and contact number, respectively. We have shown that DeepSacon can reliably predict solvent accessibility and contact number with stacked sparse autoencoder and a dropout approach.

Natural Language Video Description using Deep Recurrent Neural Networks

DTIC Science & Technology

2015-11-23

records who says what, but lacks tim- ing information. Movie scripts typically include names of all characters and most movies loosely follow the...and Jürgen Schmidhuber. A novel approach to on-line handwriting recognition based on bidirectional long short-term memory networks. In Proc. 9th Int

Dreaming of Atmospheres

NASA Astrophysics Data System (ADS)

Waldmann, Ingo

2016-10-01

Radiative transfer retrievals have become the standard in modelling of exoplanetary transmission and emission spectra. Analysing currently available observations of exoplanetary atmospheres often invoke large and correlated parameter spaces that can be difficult to map or constrain.To address these issues, we have developed the Tau-REx (tau-retrieval of exoplanets) retrieval and the RobERt spectral recognition algorithms. Tau-REx is a bayesian atmospheric retrieval framework using Nested Sampling and cluster computing to fully map these large correlated parameter spaces. Nonetheless, data volumes can become prohibitively large and we must often select a subset of potential molecular/atomic absorbers in an atmosphere.In the era of open-source, automated and self-sufficient retrieval algorithms, such manual input should be avoided. User dependent input could, in worst case scenarios, lead to incomplete models and biases in the retrieval. The RobERt algorithm is build to address these issues. RobERt is a deep belief neural (DBN) networks trained to accurately recognise molecular signatures for a wide range of planets, atmospheric thermal profiles and compositions. Using these deep neural networks, we work towards retrieval algorithms that themselves understand the nature of the observed spectra, are able to learn from current and past data and make sensible qualitative preselections of atmospheric opacities to be used for the quantitative stage of the retrieval process.In this talk I will discuss how neural networks and Bayesian Nested Sampling can be used to solve highly degenerate spectral retrieval problems and what 'dreaming' neural networks can tell us about atmospheric characteristics.

Probabilistic Models and Generative Neural Networks: Towards an Unified Framework for Modeling Normal and Impaired Neurocognitive Functions

PubMed Central

Testolin, Alberto; Zorzi, Marco

2016-01-01

Connectionist models can be characterized within the more general framework of probabilistic graphical models, which allow to efficiently describe complex statistical distributions involving a large number of interacting variables. This integration allows building more realistic computational models of cognitive functions, which more faithfully reflect the underlying neural mechanisms at the same time providing a useful bridge to higher-level descriptions in terms of Bayesian computations. Here we discuss a powerful class of graphical models that can be implemented as stochastic, generative neural networks. These models overcome many limitations associated with classic connectionist models, for example by exploiting unsupervised learning in hierarchical architectures (deep networks) and by taking into account top-down, predictive processing supported by feedback loops. We review some recent cognitive models based on generative networks, and we point out promising research directions to investigate neuropsychological disorders within this approach. Though further efforts are required in order to fill the gap between structured Bayesian models and more realistic, biophysical models of neuronal dynamics, we argue that generative neural networks have the potential to bridge these levels of analysis, thereby improving our understanding of the neural bases of cognition and of pathologies caused by brain damage. PMID:27468262

Multi-Temporal Land Cover Classification with Long Short-Term Memory Neural Networks

NASA Astrophysics Data System (ADS)

Rußwurm, M.; Körner, M.

2017-05-01

Land cover classification (LCC) is a central and wide field of research in earth observation and has already put forth a variety of classification techniques. Many approaches are based on classification techniques considering observation at certain points in time. However, some land cover classes, such as crops, change their spectral characteristics due to environmental influences and can thus not be monitored effectively with classical mono-temporal approaches. Nevertheless, these temporal observations should be utilized to benefit the classification process. After extensive research has been conducted on modeling temporal dynamics by spectro-temporal profiles using vegetation indices, we propose a deep learning approach to utilize these temporal characteristics for classification tasks. In this work, we show how long short-term memory (LSTM) neural networks can be employed for crop identification purposes with SENTINEL 2A observations from large study areas and label information provided by local authorities. We compare these temporal neural network models, i.e., LSTM and recurrent neural network (RNN), with a classical non-temporal convolutional neural network (CNN) model and an additional support vector machine (SVM) baseline. With our rather straightforward LSTM variant, we exceeded state-of-the-art classification performance, thus opening promising potential for further research.

Convolutional neural networks for event-related potential detection: impact of the architecture.

PubMed

Cecotti, H

2017-07-01

The detection of brain responses at the single-trial level in the electroencephalogram (EEG) such as event-related potentials (ERPs) is a difficult problem that requires different processing steps to extract relevant discriminant features. While most of the signal and classification techniques for the detection of brain responses are based on linear algebra, different pattern recognition techniques such as convolutional neural network (CNN), as a type of deep learning technique, have shown some interests as they are able to process the signal after limited pre-processing. In this study, we propose to investigate the performance of CNNs in relation of their architecture and in relation to how they are evaluated: a single system for each subject, or a system for all the subjects. More particularly, we want to address the change of performance that can be observed between specifying a neural network to a subject, or by considering a neural network for a group of subjects, taking advantage of a larger number of trials from different subjects. The results support the conclusion that a convolutional neural network trained on different subjects can lead to an AUC above 0.9 by using an appropriate architecture using spatial filtering and shift invariant layers.

Probabilistic Models and Generative Neural Networks: Towards an Unified Framework for Modeling Normal and Impaired Neurocognitive Functions.

PubMed

Testolin, Alberto; Zorzi, Marco

2016-01-01

Connectionist models can be characterized within the more general framework of probabilistic graphical models, which allow to efficiently describe complex statistical distributions involving a large number of interacting variables. This integration allows building more realistic computational models of cognitive functions, which more faithfully reflect the underlying neural mechanisms at the same time providing a useful bridge to higher-level descriptions in terms of Bayesian computations. Here we discuss a powerful class of graphical models that can be implemented as stochastic, generative neural networks. These models overcome many limitations associated with classic connectionist models, for example by exploiting unsupervised learning in hierarchical architectures (deep networks) and by taking into account top-down, predictive processing supported by feedback loops. We review some recent cognitive models based on generative networks, and we point out promising research directions to investigate neuropsychological disorders within this approach. Though further efforts are required in order to fill the gap between structured Bayesian models and more realistic, biophysical models of neuronal dynamics, we argue that generative neural networks have the potential to bridge these levels of analysis, thereby improving our understanding of the neural bases of cognition and of pathologies caused by brain damage.

Automated Detection of Clinically Significant Prostate Cancer in mp-MRI Images Based on an End-to-End Deep Neural Network.

PubMed

Wang, Zhiwei; Liu, Chaoyue; Cheng, Danpeng; Wang, Liang; Yang, Xin; Cheng, Kwang-Ting

2018-05-01

Automated methods for detecting clinically significant (CS) prostate cancer (PCa) in multi-parameter magnetic resonance images (mp-MRI) are of high demand. Existing methods typically employ several separate steps, each of which is optimized individually without considering the error tolerance of other steps. As a result, they could either involve unnecessary computational cost or suffer from errors accumulated over steps. In this paper, we present an automated CS PCa detection system, where all steps are optimized jointly in an end-to-end trainable deep neural network. The proposed neural network consists of concatenated subnets: 1) a novel tissue deformation network (TDN) for automated prostate detection and multimodal registration and 2) a dual-path convolutional neural network (CNN) for CS PCa detection. Three types of loss functions, i.e., classification loss, inconsistency loss, and overlap loss, are employed for optimizing all parameters of the proposed TDN and CNN. In the training phase, the two nets mutually affect each other and effectively guide registration and extraction of representative CS PCa-relevant features to achieve results with sufficient accuracy. The entire network is trained in a weakly supervised manner by providing only image-level annotations (i.e., presence/absence of PCa) without exact priors of lesions' locations. Compared with most existing systems which require supervised labels, e.g., manual delineation of PCa lesions, it is much more convenient for clinical usage. Comprehensive evaluation based on fivefold cross validation using 360 patient data demonstrates that our system achieves a high accuracy for CS PCa detection, i.e., a sensitivity of 0.6374 and 0.8978 at 0.1 and 1 false positives per normal/benign patient.

Illusory Motion Reproduced by Deep Neural Networks Trained for Prediction

PubMed Central

Watanabe, Eiji; Kitaoka, Akiyoshi; Sakamoto, Kiwako; Yasugi, Masaki; Tanaka, Kenta

2018-01-01

The cerebral cortex predicts visual motion to adapt human behavior to surrounding objects moving in real time. Although the underlying mechanisms are still unknown, predictive coding is one of the leading theories. Predictive coding assumes that the brain's internal models (which are acquired through learning) predict the visual world at all times and that errors between the prediction and the actual sensory input further refine the internal models. In the past year, deep neural networks based on predictive coding were reported for a video prediction machine called PredNet. If the theory substantially reproduces the visual information processing of the cerebral cortex, then PredNet can be expected to represent the human visual perception of motion. In this study, PredNet was trained with natural scene videos of the self-motion of the viewer, and the motion prediction ability of the obtained computer model was verified using unlearned videos. We found that the computer model accurately predicted the magnitude and direction of motion of a rotating propeller in unlearned videos. Surprisingly, it also represented the rotational motion for illusion images that were not moving physically, much like human visual perception. While the trained network accurately reproduced the direction of illusory rotation, it did not detect motion components in negative control pictures wherein people do not perceive illusory motion. This research supports the exciting idea that the mechanism assumed by the predictive coding theory is one of basis of motion illusion generation. Using sensory illusions as indicators of human perception, deep neural networks are expected to contribute significantly to the development of brain research. PMID:29599739

Illusory Motion Reproduced by Deep Neural Networks Trained for Prediction.

PubMed

Watanabe, Eiji; Kitaoka, Akiyoshi; Sakamoto, Kiwako; Yasugi, Masaki; Tanaka, Kenta

2018-01-01

The cerebral cortex predicts visual motion to adapt human behavior to surrounding objects moving in real time. Although the underlying mechanisms are still unknown, predictive coding is one of the leading theories. Predictive coding assumes that the brain's internal models (which are acquired through learning) predict the visual world at all times and that errors between the prediction and the actual sensory input further refine the internal models. In the past year, deep neural networks based on predictive coding were reported for a video prediction machine called PredNet. If the theory substantially reproduces the visual information processing of the cerebral cortex, then PredNet can be expected to represent the human visual perception of motion. In this study, PredNet was trained with natural scene videos of the self-motion of the viewer, and the motion prediction ability of the obtained computer model was verified using unlearned videos. We found that the computer model accurately predicted the magnitude and direction of motion of a rotating propeller in unlearned videos. Surprisingly, it also represented the rotational motion for illusion images that were not moving physically, much like human visual perception. While the trained network accurately reproduced the direction of illusory rotation, it did not detect motion components in negative control pictures wherein people do not perceive illusory motion. This research supports the exciting idea that the mechanism assumed by the predictive coding theory is one of basis of motion illusion generation. Using sensory illusions as indicators of human perception, deep neural networks are expected to contribute significantly to the development of brain research.

Detection of Hail Storms in Radar Imagery Using Deep Learning

NASA Technical Reports Server (NTRS)

Pullman, Melinda; Gurung, Iksha; Ramachandran, Rahul; Maskey, Manil

2017-01-01

In 2016, hail was responsible for 3.5 billion and 23 million dollars in damage to property and crops, respectively, making it the second costliest weather phenomenon in the United States. In an effort to improve hail-prediction techniques and reduce the societal impacts associated with hail storms, we propose a deep learning technique that leverages radar imagery for automatic detection of hail storms. The technique is applied to radar imagery from 2011 to 2016 for the contiguous United States and achieved a precision of 0.848. Hail storms are primarily detected through the visual interpretation of radar imagery (Mrozet al., 2017). With radars providing data every two minutes, the detection of hail storms has become a big data task. As a result, scientists have turned to neural networks that employ computer vision to identify hail-bearing storms (Marzbanet al., 2001). In this study, we propose a deep Convolutional Neural Network (ConvNet) to understand the spatial features and patterns of radar echoes for detecting hailstorms.

Deep feature learning for knee cartilage segmentation using a triplanar convolutional neural network.

PubMed

Prasoon, Adhish; Petersen, Kersten; Igel, Christian; Lauze, François; Dam, Erik; Nielsen, Mads

2013-01-01

Segmentation of anatomical structures in medical images is often based on a voxel/pixel classification approach. Deep learning systems, such as convolutional neural networks (CNNs), can infer a hierarchical representation of images that fosters categorization. We propose a novel system for voxel classification integrating three 2D CNNs, which have a one-to-one association with the xy, yz and zx planes of 3D image, respectively. We applied our method to the segmentation of tibial cartilage in low field knee MRI scans and tested it on 114 unseen scans. Although our method uses only 2D features at a single scale, it performs better than a state-of-the-art method using 3D multi-scale features. In the latter approach, the features and the classifier have been carefully adapted to the problem at hand. That we were able to get better results by a deep learning architecture that autonomously learns the features from the images is the main insight of this study.

Multimodal and Multiscale Deep Neural Networks for the Early Diagnosis of Alzheimer's Disease using structural MR and FDG-PET images.

PubMed

Lu, Donghuan; Popuri, Karteek; Ding, Gavin Weiguang; Balachandar, Rakesh; Beg, Mirza Faisal

2018-04-09

Alzheimer's Disease (AD) is a progressive neurodegenerative disease where biomarkers for disease based on pathophysiology may be able to provide objective measures for disease diagnosis and staging. Neuroimaging scans acquired from MRI and metabolism images obtained by FDG-PET provide in-vivo measurements of structure and function (glucose metabolism) in a living brain. It is hypothesized that combining multiple different image modalities providing complementary information could help improve early diagnosis of AD. In this paper, we propose a novel deep-learning-based framework to discriminate individuals with AD utilizing a multimodal and multiscale deep neural network. Our method delivers 82.4% accuracy in identifying the individuals with mild cognitive impairment (MCI) who will convert to AD at 3 years prior to conversion (86.4% combined accuracy for conversion within 1-3 years), a 94.23% sensitivity in classifying individuals with clinical diagnosis of probable AD, and a 86.3% specificity in classifying non-demented controls improving upon results in published literature.

Brain MRI analysis for Alzheimer's disease diagnosis using an ensemble system of deep convolutional neural networks.

PubMed

Islam, Jyoti; Zhang, Yanqing

2018-05-31

Alzheimer's disease is an incurable, progressive neurological brain disorder. Earlier detection of Alzheimer's disease can help with proper treatment and prevent brain tissue damage. Several statistical and machine learning models have been exploited by researchers for Alzheimer's disease diagnosis. Analyzing magnetic resonance imaging (MRI) is a common practice for Alzheimer's disease diagnosis in clinical research. Detection of Alzheimer's disease is exacting due to the similarity in Alzheimer's disease MRI data and standard healthy MRI data of older people. Recently, advanced deep learning techniques have successfully demonstrated human-level performance in numerous fields including medical image analysis. We propose a deep convolutional neural network for Alzheimer's disease diagnosis using brain MRI data analysis. While most of the existing approaches perform binary classification, our model can identify different stages of Alzheimer's disease and obtains superior performance for early-stage diagnosis. We conducted ample experiments to demonstrate that our proposed model outperformed comparative baselines on the Open Access Series of Imaging Studies dataset.

A deep convolutional neural network for recognizing foods

NASA Astrophysics Data System (ADS)

Jahani Heravi, Elnaz; Habibi Aghdam, Hamed; Puig, Domenec

2015-12-01

Controlling the food intake is an efficient way that each person can undertake to tackle the obesity problem in countries worldwide. This is achievable by developing a smartphone application that is able to recognize foods and compute their calories. State-of-art methods are chiefly based on hand-crafted feature extraction methods such as HOG and Gabor. Recent advances in large-scale object recognition datasets such as ImageNet have revealed that deep Convolutional Neural Networks (CNN) possess more representation power than the hand-crafted features. The main challenge with CNNs is to find the appropriate architecture for each problem. In this paper, we propose a deep CNN which consists of 769; 988 parameters. Our experiments show that the proposed CNN outperforms the state-of-art methods and improves the best result of traditional methods 17%. Moreover, using an ensemble of two CNNs that have been trained two different times, we are able to improve the classification performance 21:5%.

Classification of time-series images using deep convolutional neural networks

NASA Astrophysics Data System (ADS)

Hatami, Nima; Gavet, Yann; Debayle, Johan

2018-04-01

Convolutional Neural Networks (CNN) has achieved a great success in image recognition task by automatically learning a hierarchical feature representation from raw data. While the majority of Time-Series Classification (TSC) literature is focused on 1D signals, this paper uses Recurrence Plots (RP) to transform time-series into 2D texture images and then take advantage of the deep CNN classifier. Image representation of time-series introduces different feature types that are not available for 1D signals, and therefore TSC can be treated as texture image recognition task. CNN model also allows learning different levels of representations together with a classifier, jointly and automatically. Therefore, using RP and CNN in a unified framework is expected to boost the recognition rate of TSC. Experimental results on the UCR time-series classification archive demonstrate competitive accuracy of the proposed approach, compared not only to the existing deep architectures, but also to the state-of-the art TSC algorithms.

Urban Area Detection in Very High Resolution Remote Sensing Images Using Deep Convolutional Neural Networks.

PubMed

Tian, Tian; Li, Chang; Xu, Jinkang; Ma, Jiayi

2018-03-18

Detecting urban areas from very high resolution (VHR) remote sensing images plays an important role in the field of Earth observation. The recently-developed deep convolutional neural networks (DCNNs), which can extract rich features from training data automatically, have achieved outstanding performance on many image classification databases. Motivated by this fact, we propose a new urban area detection method based on DCNNs in this paper. The proposed method mainly includes three steps: (i) a visual dictionary is obtained based on the deep features extracted by pre-trained DCNNs; (ii) urban words are learned from labeled images; (iii) the urban regions are detected in a new image based on the nearest dictionary word criterion. The qualitative and quantitative experiments on different datasets demonstrate that the proposed method can obtain a remarkable overall accuracy (OA) and kappa coefficient. Moreover, it can also strike a good balance between the true positive rate (TPR) and false positive rate (FPR).

Low Data Drug Discovery with One-Shot Learning

PubMed Central

2017-01-01

Recent advances in machine learning have made significant contributions to drug discovery. Deep neural networks in particular have been demonstrated to provide significant boosts in predictive power when inferring the properties and activities of small-molecule compounds (Ma, J. et al. J. Chem. Inf. Model.2015, 55, 263–27425635324). However, the applicability of these techniques has been limited by the requirement for large amounts of training data. In this work, we demonstrate how one-shot learning can be used to significantly lower the amounts of data required to make meaningful predictions in drug discovery applications. We introduce a new architecture, the iterative refinement long short-term memory, that, when combined with graph convolutional neural networks, significantly improves learning of meaningful distance metrics over small-molecules. We open source all models introduced in this work as part of DeepChem, an open-source framework for deep-learning in drug discovery (Ramsundar, B. deepchem.io. https://github.com/deepchem/deepchem, 2016). PMID:28470045

Can we recognize horses by their ocular biometric traits using deep convolutional neural networks?

NASA Astrophysics Data System (ADS)

Trokielewicz, Mateusz; Szadkowski, Mateusz

2017-08-01

This paper aims at determining the viability of horse recognition by the means of ocular biometrics and deep convolutional neural networks (deep CNNs). Fast and accurate identification of race horses before racing is crucial for ensuring that exactly the horses that were declared are participating, using methods that are non-invasive and friendly to these delicate animals. As typical iris recognition methods require lot of fine-tuning of the method parameters and high-quality data, CNNs seem like a natural candidate to be applied for recognition thanks to their potentially excellent abilities in describing texture, combined with ease of implementation in an end-to-end manner. Also, with such approach we can easily utilize both iris and periocular features without constructing complicated algorithms for each. We thus present a simple CNN classifier, able to correctly identify almost 80% of the samples in an identification scenario, and give equal error rate (EER) of less than 10% in a verification scenario.

Robust Single Image Super-Resolution via Deep Networks With Sparse Prior.

PubMed

Liu, Ding; Wang, Zhaowen; Wen, Bihan; Yang, Jianchao; Han, Wei; Huang, Thomas S

2016-07-01

Single image super-resolution (SR) is an ill-posed problem, which tries to recover a high-resolution image from its low-resolution observation. To regularize the solution of the problem, previous methods have focused on designing good priors for natural images, such as sparse representation, or directly learning the priors from a large data set with models, such as deep neural networks. In this paper, we argue that domain expertise from the conventional sparse coding model can be combined with the key ingredients of deep learning to achieve further improved results. We demonstrate that a sparse coding model particularly designed for SR can be incarnated as a neural network with the merit of end-to-end optimization over training data. The network has a cascaded structure, which boosts the SR performance for both fixed and incremental scaling factors. The proposed training and testing schemes can be extended for robust handling of images with additional degradation, such as noise and blurring. A subjective assessment is conducted and analyzed in order to thoroughly evaluate various SR techniques. Our proposed model is tested on a wide range of images, and it significantly outperforms the existing state-of-the-art methods for various scaling factors both quantitatively and perceptually.

Enhancing SDO/HMI images using deep learning

NASA Astrophysics Data System (ADS)

Baso, C. J. Díaz; Ramos, A. Asensio

2018-06-01

Context. The Helioseismic and Magnetic Imager (HMI) provides continuum images and magnetograms with a cadence better than one per minute. It has been continuously observing the Sun 24 h a day for the past 7 yr. The trade-off between full disk observations and spatial resolution means that HMI is not adequate for analyzing the smallest-scale events in the solar atmosphere. Aims: Our aim is to develop a new method to enhance HMI data, simultaneously deconvolving and super-resolving images and magnetograms. The resulting images will mimic observations with a diffraction-limited telescope twice the diameter of HMI. Methods: Our method, which we call Enhance, is based on two deep, fully convolutional neural networks that input patches of HMI observations and output deconvolved and super-resolved data. The neural networks are trained on synthetic data obtained from simulations of the emergence of solar active regions. Results: We have obtained deconvolved and super-resolved HMI images. To solve this ill-defined problem with infinite solutions we have used a neural network approach to add prior information from the simulations. We test Enhance against Hinode data that has been degraded to a 28 cm diameter telescope showing very good consistency. The code is open source.

Automated EEG-based screening of depression using deep convolutional neural network.

PubMed

Acharya, U Rajendra; Oh, Shu Lih; Hagiwara, Yuki; Tan, Jen Hong; Adeli, Hojjat; Subha, D P

2018-07-01

In recent years, advanced neurocomputing and machine learning techniques have been used for Electroencephalogram (EEG)-based diagnosis of various neurological disorders. In this paper, a novel computer model is presented for EEG-based screening of depression using a deep neural network machine learning approach, known as Convolutional Neural Network (CNN). The proposed technique does not require a semi-manually-selected set of features to be fed into a classifier for classification. It learns automatically and adaptively from the input EEG signals to differentiate EEGs obtained from depressive and normal subjects. The model was tested using EEGs obtained from 15 normal and 15 depressed patients. The algorithm attained accuracies of 93.5% and 96.0% using EEG signals from the left and right hemisphere, respectively. It was discovered in this research that the EEG signals from the right hemisphere are more distinctive in depression than those from the left hemisphere. This discovery is consistent with recent research and revelation that the depression is associated with a hyperactive right hemisphere. An exciting extension of this research would be diagnosis of different stages and severity of depression and development of a Depression Severity Index (DSI). Copyright © 2018 Elsevier B.V. All rights reserved.

Deep recurrent neural network reveals a hierarchy of process memory during dynamic natural vision.

PubMed

Shi, Junxing; Wen, Haiguang; Zhang, Yizhen; Han, Kuan; Liu, Zhongming

2018-05-01

The human visual cortex extracts both spatial and temporal visual features to support perception and guide behavior. Deep convolutional neural networks (CNNs) provide a computational framework to model cortical representation and organization for spatial visual processing, but unable to explain how the brain processes temporal information. To overcome this limitation, we extended a CNN by adding recurrent connections to different layers of the CNN to allow spatial representations to be remembered and accumulated over time. The extended model, or the recurrent neural network (RNN), embodied a hierarchical and distributed model of process memory as an integral part of visual processing. Unlike the CNN, the RNN learned spatiotemporal features from videos to enable action recognition. The RNN better predicted cortical responses to natural movie stimuli than the CNN, at all visual areas, especially those along the dorsal stream. As a fully observable model of visual processing, the RNN also revealed a cortical hierarchy of temporal receptive window, dynamics of process memory, and spatiotemporal representations. These results support the hypothesis of process memory, and demonstrate the potential of using the RNN for in-depth computational understanding of dynamic natural vision. © 2018 Wiley Periodicals, Inc.

What the success of brain imaging implies about the neural code

PubMed Central

Guest, Olivia; Love, Bradley C

2017-01-01

The success of fMRI places constraints on the nature of the neural code. The fact that researchers can infer similarities between neural representations, despite fMRI’s limitations, implies that certain neural coding schemes are more likely than others. For fMRI to succeed given its low temporal and spatial resolution, the neural code must be smooth at the voxel and functional level such that similar stimuli engender similar internal representations. Through proof and simulation, we determine which coding schemes are plausible given both fMRI’s successes and its limitations in measuring neural activity. Deep neural network approaches, which have been forwarded as computational accounts of the ventral stream, are consistent with the success of fMRI, though functional smoothness breaks down in the later network layers. These results have implications for the nature of the neural code and ventral stream, as well as what can be successfully investigated with fMRI. DOI: http://dx.doi.org/10.7554/eLife.21397.001 PMID:28103186

Hierarchical graphical-based human pose estimation via local multi-resolution convolutional neural network

NASA Astrophysics Data System (ADS)

Zhu, Aichun; Wang, Tian; Snoussi, Hichem

2018-03-01

This paper addresses the problems of the graphical-based human pose estimation in still images, including the diversity of appearances and confounding background clutter. We present a new architecture for estimating human pose using a Convolutional Neural Network (CNN). Firstly, a Relative Mixture Deformable Model (RMDM) is defined by each pair of connected parts to compute the relative spatial information in the graphical model. Secondly, a Local Multi-Resolution Convolutional Neural Network (LMR-CNN) is proposed to train and learn the multi-scale representation of each body parts by combining different levels of part context. Thirdly, a LMR-CNN based hierarchical model is defined to explore the context information of limb parts. Finally, the experimental results demonstrate the effectiveness of the proposed deep learning approach for human pose estimation.

Hello World Deep Learning in Medical Imaging.

PubMed

Lakhani, Paras; Gray, Daniel L; Pett, Carl R; Nagy, Paul; Shih, George

2018-05-03

There is recent popularity in applying machine learning to medical imaging, notably deep learning, which has achieved state-of-the-art performance in image analysis and processing. The rapid adoption of deep learning may be attributed to the availability of machine learning frameworks and libraries to simplify their use. In this tutorial, we provide a high-level overview of how to build a deep neural network for medical image classification, and provide code that can help those new to the field begin their informatics projects.

Deep learning improves prediction of CRISPR-Cpf1 guide RNA activity.

PubMed

Kim, Hui Kwon; Min, Seonwoo; Song, Myungjae; Jung, Soobin; Choi, Jae Woo; Kim, Younggwang; Lee, Sangeun; Yoon, Sungroh; Kim, Hyongbum Henry

2018-03-01

We present two algorithms to predict the activity of AsCpf1 guide RNAs. Indel frequencies for 15,000 target sequences were used in a deep-learning framework based on a convolutional neural network to train Seq-deepCpf1. We then incorporated chromatin accessibility information to create the better-performing DeepCpf1 algorithm for cell lines for which such information is available and show that both algorithms outperform previous machine learning algorithms on our own and published data sets.

Attenuation correction for brain PET imaging using deep neural network based on dixon and ZTE MR images.

PubMed

Gong, Kuang; Yang, Jaewon; Kim, Kyungsang; El Fakhri, Georges; Seo, Youngho; Li, Quanzheng

2018-05-23

Positron Emission Tomography (PET) is a functional imaging modality widely used in neuroscience studies. To obtain meaningful quantitative results from PET images, attenuation correction is necessary during image reconstruction. For PET/MR hybrid systems, PET attenuation is challenging as Magnetic Resonance (MR) images do not reflect attenuation coefficients directly. To address this issue, we present deep neural network methods to derive the continuous attenuation coefficients for brain PET imaging from MR images. With only Dixon MR images as the network input, the existing U-net structure was adopted and analysis using forty patient data sets shows it is superior than other Dixon based methods. When both Dixon and zero echo time (ZTE) images are available, we have proposed a modified U-net structure, named GroupU-net, to efficiently make use of both Dixon and ZTE information through group convolution modules when the network goes deeper. Quantitative analysis based on fourteen real patient data sets demonstrates that both network approaches can perform better than the standard methods, and the proposed network structure can further reduce the PET quantification error compared to the U-net structure. © 2018 Institute of Physics and Engineering in Medicine.

Deep feature representation with stacked sparse auto-encoder and convolutional neural network for hyperspectral imaging-based detection of cucumber defects

USDA-ARS?s Scientific Manuscript database

It is challenging to achieve rapid and accurate processing of large amounts of hyperspectral image data. This research was aimed to develop a novel classification method by employing deep feature representation with the stacked sparse auto-encoder (SSAE) and the SSAE combined with convolutional neur...

PatterNet: a system to learn compact physical design pattern representations for pattern-based analytics

NASA Astrophysics Data System (ADS)

Lutich, Andrey

2017-07-01

This research considers the problem of generating compact vector representations of physical design patterns for analytics purposes in semiconductor patterning domain. PatterNet uses a deep artificial neural network to learn mapping of physical design patterns to a compact Euclidean hyperspace. Distances among mapped patterns in this space correspond to dissimilarities among patterns defined at the time of the network training. Once the mapping network has been trained, PatterNet embeddings can be used as feature vectors with standard machine learning algorithms, and pattern search, comparison, and clustering become trivial problems. PatterNet is inspired by the concepts developed within the framework of generative adversarial networks as well as the FaceNet. Our method facilitates a deep neural network (DNN) to learn directly the compact representation by supplying it with pairs of design patterns and dissimilarity among these patterns defined by a user. In the simplest case, the dissimilarity is represented by an area of the XOR of two patterns. Important to realize that our PatterNet approach is very different to the methods developed for deep learning on image data. In contrast to "conventional" pictures, the patterns in the CAD world are the lists of polygon vertex coordinates. The method solely relies on the promise of deep learning to discover internal structure of the incoming data and learn its hierarchical representations. Artificial intelligence arising from the combination of PatterNet and clustering analysis very precisely follows intuition of patterning/optical proximity correction experts paving the way toward human-like and human-friendly engineering tools.

High-Throughput Classification of Radiographs Using Deep Convolutional Neural Networks.

PubMed

Rajkomar, Alvin; Lingam, Sneha; Taylor, Andrew G; Blum, Michael; Mongan, John

2017-02-01

The study aimed to determine if computer vision techniques rooted in deep learning can use a small set of radiographs to perform clinically relevant image classification with high fidelity. One thousand eight hundred eighty-five chest radiographs on 909 patients obtained between January 2013 and July 2015 at our institution were retrieved and anonymized. The source images were manually annotated as frontal or lateral and randomly divided into training, validation, and test sets. Training and validation sets were augmented to over 150,000 images using standard image manipulations. We then pre-trained a series of deep convolutional networks based on the open-source GoogLeNet with various transformations of the open-source ImageNet (non-radiology) images. These trained networks were then fine-tuned using the original and augmented radiology images. The model with highest validation accuracy was applied to our institutional test set and a publicly available set. Accuracy was assessed by using the Youden Index to set a binary cutoff for frontal or lateral classification. This retrospective study was IRB approved prior to initiation. A network pre-trained on 1.2 million greyscale ImageNet images and fine-tuned on augmented radiographs was chosen. The binary classification method correctly classified 100 % (95 % CI 99.73-100 %) of both our test set and the publicly available images. Classification was rapid, at 38 images per second. A deep convolutional neural network created using non-radiological images, and an augmented set of radiographs is effective in highly accurate classification of chest radiograph view type and is a feasible, rapid method for high-throughput annotation.

Displacement back analysis for a high slope of the Dagangshan Hydroelectric Power Station based on BP neural network and particle swarm optimization.

PubMed

Liang, Zhengzhao; Gong, Bin; Tang, Chunan; Zhang, Yongbin; Ma, Tianhui

2014-01-01

The right bank high slope of the Dagangshan Hydroelectric Power Station is located in complicated geological conditions with deep fractures and unloading cracks. How to obtain the mechanical parameters and then evaluate the safety of the slope are the key problems. This paper presented a displacement back analysis for the slope using an artificial neural network model (ANN) and particle swarm optimization model (PSO). A numerical model was established to simulate the displacement increment results, acquiring training data for the artificial neural network model. The backpropagation ANN model was used to establish a mapping function between the mechanical parameters and the monitoring displacements. The PSO model was applied to initialize the weights and thresholds of the backpropagation (BP) network model and determine suitable values of the mechanical parameters. Then the elastic moduli of the rock masses were obtained according to the monitoring displacement data at different excavation stages, and the BP neural network model was proved to be valid by comparing the measured displacements, the displacements predicted by the BP neural network model, and the numerical simulation using the back-analyzed parameters. The proposed model is useful for rock mechanical parameters determination and instability investigation of rock slopes.

Displacement Back Analysis for a High Slope of the Dagangshan Hydroelectric Power Station Based on BP Neural Network and Particle Swarm Optimization

PubMed Central

Liang, Zhengzhao; Gong, Bin; Tang, Chunan; Zhang, Yongbin; Ma, Tianhui

2014-01-01

The right bank high slope of the Dagangshan Hydroelectric Power Station is located in complicated geological conditions with deep fractures and unloading cracks. How to obtain the mechanical parameters and then evaluate the safety of the slope are the key problems. This paper presented a displacement back analysis for the slope using an artificial neural network model (ANN) and particle swarm optimization model (PSO). A numerical model was established to simulate the displacement increment results, acquiring training data for the artificial neural network model. The backpropagation ANN model was used to establish a mapping function between the mechanical parameters and the monitoring displacements. The PSO model was applied to initialize the weights and thresholds of the backpropagation (BP) network model and determine suitable values of the mechanical parameters. Then the elastic moduli of the rock masses were obtained according to the monitoring displacement data at different excavation stages, and the BP neural network model was proved to be valid by comparing the measured displacements, the displacements predicted by the BP neural network model, and the numerical simulation using the back-analyzed parameters. The proposed model is useful for rock mechanical parameters determination and instability investigation of rock slopes. PMID:25140345

Computational ghost imaging using deep learning

NASA Astrophysics Data System (ADS)

Shimobaba, Tomoyoshi; Endo, Yutaka; Nishitsuji, Takashi; Takahashi, Takayuki; Nagahama, Yuki; Hasegawa, Satoki; Sano, Marie; Hirayama, Ryuji; Kakue, Takashi; Shiraki, Atsushi; Ito, Tomoyoshi

2018-04-01

Computational ghost imaging (CGI) is a single-pixel imaging technique that exploits the correlation between known random patterns and the measured intensity of light transmitted (or reflected) by an object. Although CGI can obtain two- or three-dimensional images with a single or a few bucket detectors, the quality of the reconstructed images is reduced by noise due to the reconstruction of images from random patterns. In this study, we improve the quality of CGI images using deep learning. A deep neural network is used to automatically learn the features of noise-contaminated CGI images. After training, the network is able to predict low-noise images from new noise-contaminated CGI images.

Deep Neural Networks: A New Framework for Modeling Biological Vision and Brain Information Processing.

PubMed

Kriegeskorte, Nikolaus

2015-11-24

Recent advances in neural network modeling have enabled major strides in computer vision and other artificial intelligence applications. Human-level visual recognition abilities are coming within reach of artificial systems. Artificial neural networks are inspired by the brain, and their computations could be implemented in biological neurons. Convolutional feedforward networks, which now dominate computer vision, take further inspiration from the architecture of the primate visual hierarchy. However, the current models are designed with engineering goals, not to model brain computations. Nevertheless, initial studies comparing internal representations between these models and primate brains find surprisingly similar representational spaces. With human-level performance no longer out of reach, we are entering an exciting new era, in which we will be able to build biologically faithful feedforward and recurrent computational models of how biological brains perform high-level feats of intelligence, including vision.

Change detection in multitemporal synthetic aperture radar images using dual-channel convolutional neural network

NASA Astrophysics Data System (ADS)

Liu, Tao; Li, Ying; Cao, Ying; Shen, Qiang

2017-10-01

This paper proposes a model of dual-channel convolutional neural network (CNN) that is designed for change detection in SAR images, in an effort to acquire higher detection accuracy and lower misclassification rate. This network model contains two parallel CNN channels, which can extract deep features from two multitemporal SAR images. For comparison and validation, the proposed method is tested along with other change detection algorithms on both simulated SAR images and real-world SAR images captured by different sensors. The experimental results demonstrate that the presented method outperforms the state-of-the-art techniques by a considerable margin.

Time-lagged autoencoders: Deep learning of slow collective variables for molecular kinetics

NASA Astrophysics Data System (ADS)

Wehmeyer, Christoph; Noé, Frank

2018-06-01

Inspired by the success of deep learning techniques in the physical and chemical sciences, we apply a modification of an autoencoder type deep neural network to the task of dimension reduction of molecular dynamics data. We can show that our time-lagged autoencoder reliably finds low-dimensional embeddings for high-dimensional feature spaces which capture the slow dynamics of the underlying stochastic processes—beyond the capabilities of linear dimension reduction techniques.

Classifying medical relations in clinical text via convolutional neural networks.

PubMed

He, Bin; Guan, Yi; Dai, Rui

2018-05-16

Deep learning research on relation classification has achieved solid performance in the general domain. This study proposes a convolutional neural network (CNN) architecture with a multi-pooling operation for medical relation classification on clinical records and explores a loss function with a category-level constraint matrix. Experiments using the 2010 i2b2/VA relation corpus demonstrate these models, which do not depend on any external features, outperform previous single-model methods and our best model is competitive with the existing ensemble-based method. Copyright © 2018. Published by Elsevier B.V.

Beyond Retinal Layers: A Deep Voting Model for Automated Geographic Atrophy Segmentation in SD-OCT Images

PubMed Central

Ji, Zexuan; Chen, Qiang; Niu, Sijie; Leng, Theodore; Rubin, Daniel L.

2018-01-01

Purpose To automatically and accurately segment geographic atrophy (GA) in spectral-domain optical coherence tomography (SD-OCT) images by constructing a voting system with deep neural networks without the use of retinal layer segmentation. Methods An automatic GA segmentation method for SD-OCT images based on the deep network was constructed. The structure of the deep network was composed of five layers, including one input layer, three hidden layers, and one output layer. During the training phase, the labeled A-scans with 1024 features were directly fed into the network as the input layer to obtain the deep representations. Then a soft-max classifier was trained to determine the label of each individual pixel. Finally, a voting decision strategy was used to refine the segmentation results among 10 trained models. Results Two image data sets with GA were used to evaluate the model. For the first dataset, our algorithm obtained a mean overlap ratio (OR) 86.94% ± 8.75%, absolute area difference (AAD) 11.49% ± 11.50%, and correlation coefficients (CC) 0.9857; for the second dataset, the mean OR, AAD, and CC of the proposed method were 81.66% ± 10.93%, 8.30% ± 9.09%, and 0.9952, respectively. The proposed algorithm was capable of improving over 5% and 10% segmentation accuracy, respectively, when compared with several state-of-the-art algorithms on two data sets. Conclusions Without retinal layer segmentation, the proposed algorithm could produce higher segmentation accuracy and was more stable when compared with state-of-the-art methods that relied on retinal layer segmentation results. Our model may provide reliable GA segmentations from SD-OCT images and be useful in the clinical diagnosis of advanced nonexudative AMD. Translational Relevance Based on the deep neural networks, this study presents an accurate GA segmentation method for SD-OCT images without using any retinal layer segmentation results, and may contribute to improved understanding of advanced nonexudative AMD. PMID:29302382

Beyond Retinal Layers: A Deep Voting Model for Automated Geographic Atrophy Segmentation in SD-OCT Images.

PubMed

Ji, Zexuan; Chen, Qiang; Niu, Sijie; Leng, Theodore; Rubin, Daniel L

2018-01-01

To automatically and accurately segment geographic atrophy (GA) in spectral-domain optical coherence tomography (SD-OCT) images by constructing a voting system with deep neural networks without the use of retinal layer segmentation. An automatic GA segmentation method for SD-OCT images based on the deep network was constructed. The structure of the deep network was composed of five layers, including one input layer, three hidden layers, and one output layer. During the training phase, the labeled A-scans with 1024 features were directly fed into the network as the input layer to obtain the deep representations. Then a soft-max classifier was trained to determine the label of each individual pixel. Finally, a voting decision strategy was used to refine the segmentation results among 10 trained models. Two image data sets with GA were used to evaluate the model. For the first dataset, our algorithm obtained a mean overlap ratio (OR) 86.94% ± 8.75%, absolute area difference (AAD) 11.49% ± 11.50%, and correlation coefficients (CC) 0.9857; for the second dataset, the mean OR, AAD, and CC of the proposed method were 81.66% ± 10.93%, 8.30% ± 9.09%, and 0.9952, respectively. The proposed algorithm was capable of improving over 5% and 10% segmentation accuracy, respectively, when compared with several state-of-the-art algorithms on two data sets. Without retinal layer segmentation, the proposed algorithm could produce higher segmentation accuracy and was more stable when compared with state-of-the-art methods that relied on retinal layer segmentation results. Our model may provide reliable GA segmentations from SD-OCT images and be useful in the clinical diagnosis of advanced nonexudative AMD. Based on the deep neural networks, this study presents an accurate GA segmentation method for SD-OCT images without using any retinal layer segmentation results, and may contribute to improved understanding of advanced nonexudative AMD.

Predicting backbone Cα angles and dihedrals from protein sequences by stacked sparse auto-encoder deep neural network.

PubMed

Lyons, James; Dehzangi, Abdollah; Heffernan, Rhys; Sharma, Alok; Paliwal, Kuldip; Sattar, Abdul; Zhou, Yaoqi; Yang, Yuedong

2014-10-30

Because a nearly constant distance between two neighbouring Cα atoms, local backbone structure of proteins can be represented accurately by the angle between C(αi-1)-C(αi)-C(αi+1) (θ) and a dihedral angle rotated about the C(αi)-C(αi+1) bond (τ). θ and τ angles, as the representative of structural properties of three to four amino-acid residues, offer a description of backbone conformations that is complementary to φ and ψ angles (single residue) and secondary structures (>3 residues). Here, we report the first machine-learning technique for sequence-based prediction of θ and τ angles. Predicted angles based on an independent test have a mean absolute error of 9° for θ and 34° for τ with a distribution on the θ-τ plane close to that of native values. The average root-mean-square distance of 10-residue fragment structures constructed from predicted θ and τ angles is only 1.9Å from their corresponding native structures. Predicted θ and τ angles are expected to be complementary to predicted ϕ and ψ angles and secondary structures for using in model validation and template-based as well as template-free structure prediction. The deep neural network learning technique is available as an on-line server called Structural Property prediction with Integrated DEep neuRal network (SPIDER) at http://sparks-lab.org. Copyright © 2014 Wiley Periodicals, Inc.

Deep convolutional neural network training enrichment using multi-view object-based analysis of Unmanned Aerial systems imagery for wetlands classification

NASA Astrophysics Data System (ADS)

Liu, Tao; Abd-Elrahman, Amr

2018-05-01

Deep convolutional neural network (DCNN) requires massive training datasets to trigger its image classification power, while collecting training samples for remote sensing application is usually an expensive process. When DCNN is simply implemented with traditional object-based image analysis (OBIA) for classification of Unmanned Aerial systems (UAS) orthoimage, its power may be undermined if the number training samples is relatively small. This research aims to develop a novel OBIA classification approach that can take advantage of DCNN by enriching the training dataset automatically using multi-view data. Specifically, this study introduces a Multi-View Object-based classification using Deep convolutional neural network (MODe) method to process UAS images for land cover classification. MODe conducts the classification on multi-view UAS images instead of directly on the orthoimage, and gets the final results via a voting procedure. 10-fold cross validation results show the mean overall classification accuracy increasing substantially from 65.32%, when DCNN was applied on the orthoimage to 82.08% achieved when MODe was implemented. This study also compared the performances of the support vector machine (SVM) and random forest (RF) classifiers with DCNN under traditional OBIA and the proposed multi-view OBIA frameworks. The results indicate that the advantage of DCNN over traditional classifiers in terms of accuracy is more obvious when these classifiers were applied with the proposed multi-view OBIA framework than when these classifiers were applied within the traditional OBIA framework.

Deep ensemble learning of sparse regression models for brain disease diagnosis.

PubMed

Suk, Heung-Il; Lee, Seong-Whan; Shen, Dinggang

2017-04-01

Recent studies on brain imaging analysis witnessed the core roles of machine learning techniques in computer-assisted intervention for brain disease diagnosis. Of various machine-learning techniques, sparse regression models have proved their effectiveness in handling high-dimensional data but with a small number of training samples, especially in medical problems. In the meantime, deep learning methods have been making great successes by outperforming the state-of-the-art performances in various applications. In this paper, we propose a novel framework that combines the two conceptually different methods of sparse regression and deep learning for Alzheimer's disease/mild cognitive impairment diagnosis and prognosis. Specifically, we first train multiple sparse regression models, each of which is trained with different values of a regularization control parameter. Thus, our multiple sparse regression models potentially select different feature subsets from the original feature set; thereby they have different powers to predict the response values, i.e., clinical label and clinical scores in our work. By regarding the response values from our sparse regression models as target-level representations, we then build a deep convolutional neural network for clinical decision making, which thus we call 'Deep Ensemble Sparse Regression Network.' To our best knowledge, this is the first work that combines sparse regression models with deep neural network. In our experiments with the ADNI cohort, we validated the effectiveness of the proposed method by achieving the highest diagnostic accuracies in three classification tasks. We also rigorously analyzed our results and compared with the previous studies on the ADNI cohort in the literature. Copyright © 2017 Elsevier B.V. All rights reserved.

Deep ensemble learning of sparse regression models for brain disease diagnosis

PubMed Central

Suk, Heung-Il; Lee, Seong-Whan; Shen, Dinggang

2018-01-01

Recent studies on brain imaging analysis witnessed the core roles of machine learning techniques in computer-assisted intervention for brain disease diagnosis. Of various machine-learning techniques, sparse regression models have proved their effectiveness in handling high-dimensional data but with a small number of training samples, especially in medical problems. In the meantime, deep learning methods have been making great successes by outperforming the state-of-the-art performances in various applications. In this paper, we propose a novel framework that combines the two conceptually different methods of sparse regression and deep learning for Alzheimer’s disease/mild cognitive impairment diagnosis and prognosis. Specifically, we first train multiple sparse regression models, each of which is trained with different values of a regularization control parameter. Thus, our multiple sparse regression models potentially select different feature subsets from the original feature set; thereby they have different powers to predict the response values, i.e., clinical label and clinical scores in our work. By regarding the response values from our sparse regression models as target-level representations, we then build a deep convolutional neural network for clinical decision making, which thus we call ‘ Deep Ensemble Sparse Regression Network.’ To our best knowledge, this is the first work that combines sparse regression models with deep neural network. In our experiments with the ADNI cohort, we validated the effectiveness of the proposed method by achieving the highest diagnostic accuracies in three classification tasks. We also rigorously analyzed our results and compared with the previous studies on the ADNI cohort in the literature. PMID:28167394

Deep neural network convolution (NNC) for three-class classification of diffuse lung disease opacities in high-resolution CT (HRCT): consolidation, ground-glass opacity (GGO), and normal opacity

NASA Astrophysics Data System (ADS)

Hashimoto, Noriaki; Suzuki, Kenji; Liu, Junchi; Hirano, Yasushi; MacMahon, Heber; Kido, Shoji

2018-02-01

Consolidation and ground-glass opacity (GGO) are two major types of opacities associated with diffuse lung diseases. Accurate detection and classification of such opacities are crucially important in the diagnosis of lung diseases, but the process is subjective, and suffers from interobserver variability. Our study purpose was to develop a deep neural network convolution (NNC) system for distinguishing among consolidation, GGO, and normal lung tissue in high-resolution CT (HRCT). We developed ensemble of two deep NNC models, each of which was composed of neural network regression (NNR) with an input layer, a convolution layer, a fully-connected hidden layer, and a fully-connected output layer followed by a thresholding layer. The output layer of each NNC provided a map for the likelihood of being each corresponding lung opacity of interest. The two NNC models in the ensemble were connected in a class-selection layer. We trained our NNC ensemble with pairs of input 2D axial slices and "teaching" probability maps for the corresponding lung opacity, which were obtained by combining three radiologists' annotations. We randomly selected 10 and 40 slices from HRCT scans of 172 patients for each class as a training and test set, respectively. Our NNC ensemble achieved an area under the receiver-operating-characteristic (ROC) curve (AUC) of 0.981 and 0.958 in distinction of consolidation and GGO, respectively, from normal opacity, yielding a classification accuracy of 93.3% among 3 classes. Thus, our deep-NNC-based system for classifying diffuse lung diseases achieved high accuracies for classification of consolidation, GGO, and normal opacity.

Study on the Classification of GAOFEN-3 Polarimetric SAR Images Using Deep Neural Network

NASA Astrophysics Data System (ADS)

Zhang, J.; Zhang, J.; Zhao, Z.

2018-04-01

Polarimetric Synthetic Aperture Radar (POLSAR) imaging principle determines that the image quality will be affected by speckle noise. So the recognition accuracy of traditional image classification methods will be reduced by the effect of this interference. Since the date of submission, Deep Convolutional Neural Network impacts on the traditional image processing methods and brings the field of computer vision to a new stage with the advantages of a strong ability to learn deep features and excellent ability to fit large datasets. Based on the basic characteristics of polarimetric SAR images, the paper studied the types of the surface cover by using the method of Deep Learning. We used the fully polarimetric SAR features of different scales to fuse RGB images to the GoogLeNet model based on convolution neural network Iterative training, and then use the trained model to test the classification of data validation.First of all, referring to the optical image, we mark the surface coverage type of GF-3 POLSAR image with 8m resolution, and then collect the samples according to different categories. To meet the GoogLeNet model requirements of 256 × 256 pixel image input and taking into account the lack of full-resolution SAR resolution, the original image should be pre-processed in the process of resampling. In this paper, POLSAR image slice samples of different scales with sampling intervals of 2 m and 1 m to be trained separately and validated by the verification dataset. Among them, the training accuracy of GoogLeNet model trained with resampled 2-m polarimetric SAR image is 94.89 %, and that of the trained SAR image with resampled 1 m is 92.65 %.

Combining deep residual neural network features with supervised machine learning algorithms to classify diverse food image datasets.

PubMed

McAllister, Patrick; Zheng, Huiru; Bond, Raymond; Moorhead, Anne

2018-04-01

Obesity is increasing worldwide and can cause many chronic conditions such as type-2 diabetes, heart disease, sleep apnea, and some cancers. Monitoring dietary intake through food logging is a key method to maintain a healthy lifestyle to prevent and manage obesity. Computer vision methods have been applied to food logging to automate image classification for monitoring dietary intake. In this work we applied pretrained ResNet-152 and GoogleNet convolutional neural networks (CNNs), initially trained using ImageNet Large Scale Visual Recognition Challenge (ILSVRC) dataset with MatConvNet package, to extract features from food image datasets; Food 5K, Food-11, RawFooT-DB, and Food-101. Deep features were extracted from CNNs and used to train machine learning classifiers including artificial neural network (ANN), support vector machine (SVM), Random Forest, and Naive Bayes. Results show that using ResNet-152 deep features with SVM with RBF kernel can accurately detect food items with 99.4% accuracy using Food-5K validation food image dataset and 98.8% with Food-5K evaluation dataset using ANN, SVM-RBF, and Random Forest classifiers. Trained with ResNet-152 features, ANN can achieve 91.34%, 99.28% when applied to Food-11 and RawFooT-DB food image datasets respectively and SVM with RBF kernel can achieve 64.98% with Food-101 image dataset. From this research it is clear that using deep CNN features can be used efficiently for diverse food item image classification. The work presented in this research shows that pretrained ResNet-152 features provide sufficient generalisation power when applied to a range of food image classification tasks. Copyright © 2018 Elsevier Ltd. All rights reserved.

Neonatal Seizure Detection Using Deep Convolutional Neural Networks.

PubMed

Ansari, Amir H; Cherian, Perumpillichira J; Caicedo, Alexander; Naulaers, Gunnar; De Vos, Maarten; Van Huffel, Sabine

2018-04-02

Identifying a core set of features is one of the most important steps in the development of an automated seizure detector. In most of the published studies describing features and seizure classifiers, the features were hand-engineered, which may not be optimal. The main goal of the present paper is using deep convolutional neural networks (CNNs) and random forest to automatically optimize feature selection and classification. The input of the proposed classifier is raw multi-channel EEG and the output is the class label: seizure/nonseizure. By training this network, the required features are optimized, while fitting a nonlinear classifier on the features. After training the network with EEG recordings of 26 neonates, five end layers performing the classification were replaced with a random forest classifier in order to improve the performance. This resulted in a false alarm rate of 0.9 per hour and seizure detection rate of 77% using a test set of EEG recordings of 22 neonates that also included dubious seizures. The newly proposed CNN classifier outperformed three data-driven feature-based approaches and performed similar to a previously developed heuristic method.

Conditional random field modelling of interactions between findings in mammography

NASA Astrophysics Data System (ADS)

Kooi, Thijs; Mordang, Jan-Jurre; Karssemeijer, Nico

2017-03-01

Recent breakthroughs in training deep neural network architectures, in particular deep Convolutional Neural Networks (CNNs), made a big impact on vision research and are increasingly responsible for advances in Computer Aided Diagnosis (CAD). Since many natural scenes and medical images vary in size and are too large to feed to the networks as a whole, two stage systems are typically employed, where in the first stage, small regions of interest in the image are located and presented to the network as training and test data. These systems allow us to harness accurate region based annotations, making the problem easier to learn. However, information is processed purely locally and context is not taken into account. In this paper, we present preliminary work on the employment of a Conditional Random Field (CRF) that is trained on top the CNN to model contextual interactions such as the presence of other suspicious regions, for mammography CAD. The model can easily be extended to incorporate other sources of information, such as symmetry, temporal change and various patient covariates and is general in the sense that it can have application in other CAD problems.

Location Sensitive Deep Convolutional Neural Networks for Segmentation of White Matter Hyperintensities.

PubMed

Ghafoorian, Mohsen; Karssemeijer, Nico; Heskes, Tom; van Uden, Inge W M; Sanchez, Clara I; Litjens, Geert; de Leeuw, Frank-Erik; van Ginneken, Bram; Marchiori, Elena; Platel, Bram

2017-07-11

The anatomical location of imaging features is of crucial importance for accurate diagnosis in many medical tasks. Convolutional neural networks (CNN) have had huge successes in computer vision, but they lack the natural ability to incorporate the anatomical location in their decision making process, hindering success in some medical image analysis tasks. In this paper, to integrate the anatomical location information into the network, we propose several deep CNN architectures that consider multi-scale patches or take explicit location features while training. We apply and compare the proposed architectures for segmentation of white matter hyperintensities in brain MR images on a large dataset. As a result, we observe that the CNNs that incorporate location information substantially outperform a conventional segmentation method with handcrafted features as well as CNNs that do not integrate location information. On a test set of 50 scans, the best configuration of our networks obtained a Dice score of 0.792, compared to 0.805 for an independent human observer. Performance levels of the machine and the independent human observer were not statistically significantly different (p-value = 0.06).

Developing convolutional neural networks for measuring climate change opinions from social media data

NASA Astrophysics Data System (ADS)

Mao, H.; Bhaduri, B. L.

2016-12-01

Understanding public opinions on climate change is important for policy making. Public opinion, however, is typically measured with national surveys, which are often too expensive and thus being updated at a low frequency. Twitter has become a major platform for people to express their opinions on social and political issues. Our work attempts to understand if Twitter data can provide complimentary insights about climate change perceptions. Since the nature of social media is real-time, this data source can especially help us understand how public opinion changes over time in response to climate events and hazards, which though is very difficult to be captured by manual surveys. We use the Twitter Streaming API to collect tweets that contain keywords, "climate change" or "#climatechange". Traditional machine-learning based opinion mining algorithms require a significant amount of labeled data. Data labeling is notoriously time consuming. To address this problem, we use hashtags (a significant feature used to mark topics of tweets) to annotate tweets automatically. For example, hashtags, #climatedenial and #climatescam, are negative opinion labels, while #actonclimate and #climateaction are positive. Following this method, we can obtain a large amount of training data without human labor. This labeled dataset is used to train a deep convolutional neural network that classifies tweets into positive (i.e. believe in climate change) and negative (i.e. do not believe). Based on the positive/negative tweets obtained, we will further analyze risk perceptions and opinions towards policy support. In addition, we analyze twitter user profiles to understand the demographics of proponents and opponents of climate change. Deep learning techniques, especially convolutional deep neural networks, have achieved much success in computer vision. In this work, we propose a convolutional neural network architecture for understanding opinions within text. This method is compared with lexicon-based opinion analysis approaches. Results and the advantages/limitations of this method are to be discussed.

Prolongation of SMAP to Spatiotemporally Seamless Coverage of Continental U.S. Using a Deep Learning Neural Network

NASA Astrophysics Data System (ADS)

Fang, Kuai; Shen, Chaopeng; Kifer, Daniel; Yang, Xiao

2017-11-01

The Soil Moisture Active Passive (SMAP) mission has delivered valuable sensing of surface soil moisture since 2015. However, it has a short time span and irregular revisit schedules. Utilizing a state-of-the-art time series deep learning neural network, Long Short-Term Memory (LSTM), we created a system that predicts SMAP level-3 moisture product with atmospheric forcings, model-simulated moisture, and static physiographic attributes as inputs. The system removes most of the bias with model simulations and improves predicted moisture climatology, achieving small test root-mean-square errors (<0.035) and high-correlation coefficients >0.87 for over 75% of Continental United States, including the forested southeast. As the first application of LSTM in hydrology, we show the proposed network avoids overfitting and is robust for both temporal and spatial extrapolation tests. LSTM generalizes well across regions with distinct climates and environmental settings. With high fidelity to SMAP, LSTM shows great potential for hindcasting, data assimilation, and weather forecasting.

Dreaming of Atmospheres

NASA Astrophysics Data System (ADS)

Waldmann, I. P.

2016-04-01

Here, we introduce the RobERt (Robotic Exoplanet Recognition) algorithm for the classification of exoplanetary emission spectra. Spectral retrieval of exoplanetary atmospheres frequently requires the preselection of molecular/atomic opacities to be defined by the user. In the era of open-source, automated, and self-sufficient retrieval algorithms, manual input should be avoided. User dependent input could, in worst-case scenarios, lead to incomplete models and biases in the retrieval. The RobERt algorithm is based on deep-belief neural (DBN) networks trained to accurately recognize molecular signatures for a wide range of planets, atmospheric thermal profiles, and compositions. Reconstructions of the learned features, also referred to as the “dreams” of the network, indicate good convergence and an accurate representation of molecular features in the DBN. Using these deep neural networks, we work toward retrieval algorithms that themselves understand the nature of the observed spectra, are able to learn from current and past data, and make sensible qualitative preselections of atmospheric opacities to be used for the quantitative stage of the retrieval process.

A hybrid technique for speech segregation and classification using a sophisticated deep neural network

PubMed Central

Nawaz, Tabassam; Mehmood, Zahid; Rashid, Muhammad; Habib, Hafiz Adnan

2018-01-01

Recent research on speech segregation and music fingerprinting has led to improvements in speech segregation and music identification algorithms. Speech and music segregation generally involves the identification of music followed by speech segregation. However, music segregation becomes a challenging task in the presence of noise. This paper proposes a novel method of speech segregation for unlabelled stationary noisy audio signals using the deep belief network (DBN) model. The proposed method successfully segregates a music signal from noisy audio streams. A recurrent neural network (RNN)-based hidden layer segregation model is applied to remove stationary noise. Dictionary-based fisher algorithms are employed for speech classification. The proposed method is tested on three datasets (TIMIT, MIR-1K, and MusicBrainz), and the results indicate the robustness of proposed method for speech segregation. The qualitative and quantitative analysis carried out on three datasets demonstrate the efficiency of the proposed method compared to the state-of-the-art speech segregation and classification-based methods. PMID:29558485

Analyzing Distributional Learning of Phonemic Categories in Unsupervised Deep Neural Networks

PubMed Central

Räsänen, Okko; Nagamine, Tasha; Mesgarani, Nima

2017-01-01

Infants’ speech perception adapts to the phonemic categories of their native language, a process assumed to be driven by the distributional properties of speech. This study investigates whether deep neural networks (DNNs), the current state-of-the-art in distributional feature learning, are capable of learning phoneme-like representations of speech in an unsupervised manner. We trained DNNs with unlabeled and labeled speech and analyzed the activations of each layer with respect to the phones in the input segments. The analyses reveal that the emergence of phonemic invariance in DNNs is dependent on the availability of phonemic labeling of the input during the training. No increased phonemic selectivity of the hidden layers was observed in the purely unsupervised networks despite successful learning of low-dimensional representations for speech. This suggests that additional learning constraints or more sophisticated models are needed to account for the emergence of phone-like categories in distributional learning operating on natural speech. PMID:29359204

Uncertainties in Parameters Estimated with Neural Networks: Application to Strong Gravitational Lensing

DOE PAGES

Perreault Levasseur, Laurence; Hezaveh, Yashar D.; Wechsler, Risa H.

2017-11-15

In Hezaveh et al. (2017) we showed that deep learning can be used for model parameter estimation and trained convolutional neural networks to determine the parameters of strong gravitational lensing systems. Here we demonstrate a method for obtaining the uncertainties of these parameters. We review the framework of variational inference to obtain approximate posteriors of Bayesian neural networks and apply it to a network trained to estimate the parameters of the Singular Isothermal Ellipsoid plus external shear and total flux magnification. We show that the method can capture the uncertainties due to different levels of noise in the input data,more » as well as training and architecture-related errors made by the network. To evaluate the accuracy of the resulting uncertainties, we calculate the coverage probabilities of marginalized distributions for each lensing parameter. By tuning a single hyperparameter, the dropout rate, we obtain coverage probabilities approximately equal to the confidence levels for which they were calculated, resulting in accurate and precise uncertainty estimates. Our results suggest that neural networks can be a fast alternative to Monte Carlo Markov Chains for parameter uncertainty estimation in many practical applications, allowing more than seven orders of magnitude improvement in speed.« less

Uncertainties in Parameters Estimated with Neural Networks: Application to Strong Gravitational Lensing

NASA Astrophysics Data System (ADS)

Perreault Levasseur, Laurence; Hezaveh, Yashar D.; Wechsler, Risa H.

2017-11-01

In Hezaveh et al. we showed that deep learning can be used for model parameter estimation and trained convolutional neural networks to determine the parameters of strong gravitational-lensing systems. Here we demonstrate a method for obtaining the uncertainties of these parameters. We review the framework of variational inference to obtain approximate posteriors of Bayesian neural networks and apply it to a network trained to estimate the parameters of the Singular Isothermal Ellipsoid plus external shear and total flux magnification. We show that the method can capture the uncertainties due to different levels of noise in the input data, as well as training and architecture-related errors made by the network. To evaluate the accuracy of the resulting uncertainties, we calculate the coverage probabilities of marginalized distributions for each lensing parameter. By tuning a single variational parameter, the dropout rate, we obtain coverage probabilities approximately equal to the confidence levels for which they were calculated, resulting in accurate and precise uncertainty estimates. Our results suggest that the application of approximate Bayesian neural networks to astrophysical modeling problems can be a fast alternative to Monte Carlo Markov Chains, allowing orders of magnitude improvement in speed.

Uncertainties in Parameters Estimated with Neural Networks: Application to Strong Gravitational Lensing

DOE Office of Scientific and Technical Information (OSTI.GOV)

Perreault Levasseur, Laurence; Hezaveh, Yashar D.; Wechsler, Risa H.

In Hezaveh et al. (2017) we showed that deep learning can be used for model parameter estimation and trained convolutional neural networks to determine the parameters of strong gravitational lensing systems. Here we demonstrate a method for obtaining the uncertainties of these parameters. We review the framework of variational inference to obtain approximate posteriors of Bayesian neural networks and apply it to a network trained to estimate the parameters of the Singular Isothermal Ellipsoid plus external shear and total flux magnification. We show that the method can capture the uncertainties due to different levels of noise in the input data,more » as well as training and architecture-related errors made by the network. To evaluate the accuracy of the resulting uncertainties, we calculate the coverage probabilities of marginalized distributions for each lensing parameter. By tuning a single hyperparameter, the dropout rate, we obtain coverage probabilities approximately equal to the confidence levels for which they were calculated, resulting in accurate and precise uncertainty estimates. Our results suggest that neural networks can be a fast alternative to Monte Carlo Markov Chains for parameter uncertainty estimation in many practical applications, allowing more than seven orders of magnitude improvement in speed.« less

Enhanced Higgs boson to τ(+)τ(-) search with deep learning.

PubMed

Baldi, P; Sadowski, P; Whiteson, D

2015-03-20

The Higgs boson is thought to provide the interaction that imparts mass to the fundamental fermions, but while measurements at the Large Hadron Collider (LHC) are consistent with this hypothesis, current analysis techniques lack the statistical power to cross the traditional 5σ significance barrier without more data. Deep learning techniques have the potential to increase the statistical power of this analysis by automatically learning complex, high-level data representations. In this work, deep neural networks are used to detect the decay of the Higgs boson to a pair of tau leptons. A Bayesian optimization algorithm is used to tune the network architecture and training algorithm hyperparameters, resulting in a deep network of eight nonlinear processing layers that improves upon the performance of shallow classifiers even without the use of features specifically engineered by physicists for this application. The improvement in discovery significance is equivalent to an increase in the accumulated data set of 25%.

Fully automatic detection and segmentation of abdominal aortic thrombus in post-operative CTA images using Deep Convolutional Neural Networks.

PubMed

López-Linares, Karen; Aranjuelo, Nerea; Kabongo, Luis; Maclair, Gregory; Lete, Nerea; Ceresa, Mario; García-Familiar, Ainhoa; Macía, Iván; González Ballester, Miguel A

2018-05-01

Computerized Tomography Angiography (CTA) based follow-up of Abdominal Aortic Aneurysms (AAA) treated with Endovascular Aneurysm Repair (EVAR) is essential to evaluate the progress of the patient and detect complications. In this context, accurate quantification of post-operative thrombus volume is required. However, a proper evaluation is hindered by the lack of automatic, robust and reproducible thrombus segmentation algorithms. We propose a new fully automatic approach based on Deep Convolutional Neural Networks (DCNN) for robust and reproducible thrombus region of interest detection and subsequent fine thrombus segmentation. The DetecNet detection network is adapted to perform region of interest extraction from a complete CTA and a new segmentation network architecture, based on Fully Convolutional Networks and a Holistically-Nested Edge Detection Network, is presented. These networks are trained, validated and tested in 13 post-operative CTA volumes of different patients using a 4-fold cross-validation approach to provide more robustness to the results. Our pipeline achieves a Dice score of more than 82% for post-operative thrombus segmentation and provides a mean relative volume difference between ground truth and automatic segmentation that lays within the experienced human observer variance without the need of human intervention in most common cases. Copyright © 2018 Elsevier B.V. All rights reserved.

Semi-Supervised Recurrent Neural Network for Adverse Drug Reaction mention extraction.

PubMed

Gupta, Shashank; Pawar, Sachin; Ramrakhiyani, Nitin; Palshikar, Girish Keshav; Varma, Vasudeva

2018-06-13

Social media is a useful platform to share health-related information due to its vast reach. This makes it a good candidate for public-health monitoring tasks, specifically for pharmacovigilance. We study the problem of extraction of Adverse-Drug-Reaction (ADR) mentions from social media, particularly from Twitter. Medical information extraction from social media is challenging, mainly due to short and highly informal nature of text, as compared to more technical and formal medical reports. Current methods in ADR mention extraction rely on supervised learning methods, which suffer from labeled data scarcity problem. The state-of-the-art method uses deep neural networks, specifically a class of Recurrent Neural Network (RNN) which is Long-Short-Term-Memory network (LSTM). Deep neural networks, due to their large number of free parameters rely heavily on large annotated corpora for learning the end task. But in the real-world, it is hard to get large labeled data, mainly due to the heavy cost associated with the manual annotation. To this end, we propose a novel semi-supervised learning based RNN model, which can leverage unlabeled data also present in abundance on social media. Through experiments we demonstrate the effectiveness of our method, achieving state-of-the-art performance in ADR mention extraction. In this study, we tackle the problem of labeled data scarcity for Adverse Drug Reaction mention extraction from social media and propose a novel semi-supervised learning based method which can leverage large unlabeled corpus available in abundance on the web. Through empirical study, we demonstrate that our proposed method outperforms fully supervised learning based baseline which relies on large manually annotated corpus for a good performance.

Comparing deep neural network and other machine learning algorithms for stroke prediction in a large-scale population-based electronic medical claims database.

PubMed

Chen-Ying Hung; Wei-Chen Chen; Po-Tsun Lai; Ching-Heng Lin; Chi-Chun Lee

2017-07-01

Electronic medical claims (EMCs) can be used to accurately predict the occurrence of a variety of diseases, which can contribute to precise medical interventions. While there is a growing interest in the application of machine learning (ML) techniques to address clinical problems, the use of deep-learning in healthcare have just gained attention recently. Deep learning, such as deep neural network (DNN), has achieved impressive results in the areas of speech recognition, computer vision, and natural language processing in recent years. However, deep learning is often difficult to comprehend due to the complexities in its framework. Furthermore, this method has not yet been demonstrated to achieve a better performance comparing to other conventional ML algorithms in disease prediction tasks using EMCs. In this study, we utilize a large population-based EMC database of around 800,000 patients to compare DNN with three other ML approaches for predicting 5-year stroke occurrence. The result shows that DNN and gradient boosting decision tree (GBDT) can result in similarly high prediction accuracies that are better compared to logistic regression (LR) and support vector machine (SVM) approaches. Meanwhile, DNN achieves optimal results by using lesser amounts of patient data when comparing to GBDT method.

Protein Solvent-Accessibility Prediction by a Stacked Deep Bidirectional Recurrent Neural Network.

PubMed

Zhang, Buzhong; Li, Linqing; Lü, Qiang

2018-05-25

Residue solvent accessibility is closely related to the spatial arrangement and packing of residues. Predicting the solvent accessibility of a protein is an important step to understand its structure and function. In this work, we present a deep learning method to predict residue solvent accessibility, which is based on a stacked deep bidirectional recurrent neural network applied to sequence profiles. To capture more long-range sequence information, a merging operator was proposed when bidirectional information from hidden nodes was merged for outputs. Three types of merging operators were used in our improved model, with a long short-term memory network performing as a hidden computing node. The trained database was constructed from 7361 proteins extracted from the PISCES server using a cut-off of 25% sequence identity. Sequence-derived features including position-specific scoring matrix, physical properties, physicochemical characteristics, conservation score and protein coding were used to represent a residue. Using this method, predictive values of continuous relative solvent-accessible area were obtained, and then, these values were transformed into binary states with predefined thresholds. Our experimental results showed that our deep learning method improved prediction quality relative to current methods, with mean absolute error and Pearson's correlation coefficient values of 8.8% and 74.8%, respectively, on the CB502 dataset and 8.2% and 78%, respectively, on the Manesh215 dataset.

Identifying Corresponding Patches in SAR and Optical Images With a Pseudo-Siamese CNN

NASA Astrophysics Data System (ADS)

Hughes, Lloyd H.; Schmitt, Michael; Mou, Lichao; Wang, Yuanyuan; Zhu, Xiao Xiang

2018-05-01

In this letter, we propose a pseudo-siamese convolutional neural network (CNN) architecture that enables to solve the task of identifying corresponding patches in very-high-resolution (VHR) optical and synthetic aperture radar (SAR) remote sensing imagery. Using eight convolutional layers each in two parallel network streams, a fully connected layer for the fusion of the features learned in each stream, and a loss function based on binary cross-entropy, we achieve a one-hot indication if two patches correspond or not. The network is trained and tested on an automatically generated dataset that is based on a deterministic alignment of SAR and optical imagery via previously reconstructed and subsequently co-registered 3D point clouds. The satellite images, from which the patches comprising our dataset are extracted, show a complex urban scene containing many elevated objects (i.e. buildings), thus providing one of the most difficult experimental environments. The achieved results show that the network is able to predict corresponding patches with high accuracy, thus indicating great potential for further development towards a generalized multi-sensor key-point matching procedure. Index Terms-synthetic aperture radar (SAR), optical imagery, data fusion, deep learning, convolutional neural networks (CNN), image matching, deep matching

Using YOLO based deep learning network for real time detection and localization of lung nodules from low dose CT scans

NASA Astrophysics Data System (ADS)

Ramachandran S., Sindhu; George, Jose; Skaria, Shibon; V. V., Varun

2018-02-01

Lung cancer is the leading cause of cancer related deaths in the world. The survival rate can be improved if the presence of lung nodules are detected early. This has also led to more focus being given to computer aided detection (CAD) and diagnosis of lung nodules. The arbitrariness of shape, size and texture of lung nodules is a challenge to be faced when developing these detection systems. In the proposed work we use convolutional neural networks to learn the features for nodule detection, replacing the traditional method of handcrafting features like geometric shape or texture. Our network uses the DetectNet architecture based on YOLO (You Only Look Once) to detect the nodules in CT scans of lung. In this architecture, object detection is treated as a regression problem with a single convolutional network simultaneously predicting multiple bounding boxes and class probabilities for those boxes. By performing training using chest CT scans from Lung Image Database Consortium (LIDC), NVIDIA DIGITS and Caffe deep learning framework, we show that nodule detection using this single neural network can result in reasonably low false positive rates with high sensitivity and precision.

Deep-HiTS: Rotation Invariant Convolutional Neural Network for Transient Detection

NASA Astrophysics Data System (ADS)

Cabrera-Vives, Guillermo; Reyes, Ignacio; Förster, Francisco; Estévez, Pablo A.; Maureira, Juan-Carlos

2017-02-01

We introduce Deep-HiTS, a rotation-invariant convolutional neural network (CNN) model for classifying images of transient candidates into artifacts or real sources for the High cadence Transient Survey (HiTS). CNNs have the advantage of learning the features automatically from the data while achieving high performance. We compare our CNN model against a feature engineering approach using random forests (RFs). We show that our CNN significantly outperforms the RF model, reducing the error by almost half. Furthermore, for a fixed number of approximately 2000 allowed false transient candidates per night, we are able to reduce the misclassified real transients by approximately one-fifth. To the best of our knowledge, this is the first time CNNs have been used to detect astronomical transient events. Our approach will be very useful when processing images from next generation instruments such as the Large Synoptic Survey Telescope. We have made all our code and data available to the community for the sake of allowing further developments and comparisons at https://github.com/guille-c/Deep-HiTS. Deep-HiTS is licensed under the terms of the GNU General Public License v3.0.

A Deep Learning based Approach to Reduced Order Modeling of Fluids using LSTM Neural Networks

NASA Astrophysics Data System (ADS)

Mohan, Arvind; Gaitonde, Datta

2017-11-01

Reduced Order Modeling (ROM) can be used as surrogates to prohibitively expensive simulations to model flow behavior for long time periods. ROM is predicated on extracting dominant spatio-temporal features of the flow from CFD or experimental datasets. We explore ROM development with a deep learning approach, which comprises of learning functional relationships between different variables in large datasets for predictive modeling. Although deep learning and related artificial intelligence based predictive modeling techniques have shown varied success in other fields, such approaches are in their initial stages of application to fluid dynamics. Here, we explore the application of the Long Short Term Memory (LSTM) neural network to sequential data, specifically to predict the time coefficients of Proper Orthogonal Decomposition (POD) modes of the flow for future timesteps, by training it on data at previous timesteps. The approach is demonstrated by constructing ROMs of several canonical flows. Additionally, we show that statistical estimates of stationarity in the training data can indicate a priori how amenable a given flow-field is to this approach. Finally, the potential and limitations of deep learning based ROM approaches will be elucidated and further developments discussed.

Ship detection leveraging deep neural networks in WorldView-2 images

NASA Astrophysics Data System (ADS)

Yamamoto, T.; Kazama, Y.

2017-10-01

Interpretation of high-resolution satellite images has been so difficult that skilled interpreters must have checked the satellite images manually because of the following issues. One is the requirement of the high detection accuracy rate. The other is the variety of the target, taking ships for example, there are many kinds of ships, such as boat, cruise ship, cargo ship, aircraft carrier, and so on. Furthermore, there are similar appearance objects throughout the image; therefore, it is often difficult even for the skilled interpreters to distinguish what object the pixels really compose. In this paper, we explore the feasibility of object extraction leveraging deep learning with high-resolution satellite images, especially focusing on ship detection. We calculated the detection accuracy using the WorldView-2 images. First, we collected the training images labelled as "ship" and "not ship". After preparing the training data, we defined the deep neural network model to judge whether ships are existing or not, and trained them with about 50,000 training images for each label. Subsequently, we scanned the evaluation image with different resolution windows and extracted the "ship" images. Experimental result shows the effectiveness of the deep learning based object detection.

Inner and Outer Recursive Neural Networks for Chemoinformatics Applications.

PubMed

Urban, Gregor; Subrahmanya, Niranjan; Baldi, Pierre

2018-02-26

Deep learning methods applied to problems in chemoinformatics often require the use of recursive neural networks to handle data with graphical structure and variable size. We present a useful classification of recursive neural network approaches into two classes, the inner and outer approach. The inner approach uses recursion inside the underlying graph, to essentially "crawl" the edges of the graph, while the outer approach uses recursion outside the underlying graph, to aggregate information over progressively longer distances in an orthogonal direction. We illustrate the inner and outer approaches on several examples. More importantly, we provide open-source implementations [available at www.github.com/Chemoinformatics/InnerOuterRNN and cdb.ics.uci.edu ] for both approaches in Tensorflow which can be used in combination with training data to produce efficient models for predicting the physical, chemical, and biological properties of small molecules.

Vanishing points detection using combination of fast Hough transform and deep learning

NASA Astrophysics Data System (ADS)

Sheshkus, Alexander; Ingacheva, Anastasia; Nikolaev, Dmitry

2018-04-01

In this paper we propose a novel method for vanishing points detection based on convolutional neural network (CNN) approach and fast Hough transform algorithm. We show how to determine fast Hough transform neural network layer and how to use it in order to increase usability of the neural network approach to the vanishing point detection task. Our algorithm includes CNN with consequence of convolutional and fast Hough transform layers. We are building estimator for distribution of possible vanishing points in the image. This distribution can be used to find candidates of vanishing point. We provide experimental results from tests of suggested method using images collected from videos of road trips. Our approach shows stable result on test images with different projective distortions and noise. Described approach can be effectively implemented for mobile GPU and CPU.

Single Image Super-Resolution Based on Multi-Scale Competitive Convolutional Neural Network

PubMed Central

Qu, Xiaobo; He, Yifan

2018-01-01

Deep convolutional neural networks (CNNs) are successful in single-image super-resolution. Traditional CNNs are limited to exploit multi-scale contextual information for image reconstruction due to the fixed convolutional kernel in their building modules. To restore various scales of image details, we enhance the multi-scale inference capability of CNNs by introducing competition among multi-scale convolutional filters, and build up a shallow network under limited computational resources. The proposed network has the following two advantages: (1) the multi-scale convolutional kernel provides the multi-context for image super-resolution, and (2) the maximum competitive strategy adaptively chooses the optimal scale of information for image reconstruction. Our experimental results on image super-resolution show that the performance of the proposed network outperforms the state-of-the-art methods. PMID:29509666

Single Image Super-Resolution Based on Multi-Scale Competitive Convolutional Neural Network.

PubMed

Du, Xiaofeng; Qu, Xiaobo; He, Yifan; Guo, Di

2018-03-06

Deep convolutional neural networks (CNNs) are successful in single-image super-resolution. Traditional CNNs are limited to exploit multi-scale contextual information for image reconstruction due to the fixed convolutional kernel in their building modules. To restore various scales of image details, we enhance the multi-scale inference capability of CNNs by introducing competition among multi-scale convolutional filters, and build up a shallow network under limited computational resources. The proposed network has the following two advantages: (1) the multi-scale convolutional kernel provides the multi-context for image super-resolution, and (2) the maximum competitive strategy adaptively chooses the optimal scale of information for image reconstruction. Our experimental results on image super-resolution show that the performance of the proposed network outperforms the state-of-the-art methods.

Evaluation of Deep Learning Representations of Spatial Storm Data

NASA Astrophysics Data System (ADS)

Gagne, D. J., II; Haupt, S. E.; Nychka, D. W.

2017-12-01

The spatial structure of a severe thunderstorm and its surrounding environment provide useful information about the potential for severe weather hazards, including tornadoes, hail, and high winds. Statistics computed over the area of a storm or from the pre-storm environment can provide descriptive information but fail to capture structural information. Because the storm environment is a complex, high-dimensional space, identifying methods to encode important spatial storm information in a low-dimensional form should aid analysis and prediction of storms by statistical and machine learning models. Principal component analysis (PCA), a more traditional approach, transforms high-dimensional data into a set of linearly uncorrelated, orthogonal components ordered by the amount of variance explained by each component. The burgeoning field of deep learning offers two potential approaches to this problem. Convolutional Neural Networks are a supervised learning method for transforming spatial data into a hierarchical set of feature maps that correspond with relevant combinations of spatial structures in the data. Generative Adversarial Networks (GANs) are an unsupervised deep learning model that uses two neural networks trained against each other to produce encoded representations of spatial data. These different spatial encoding methods were evaluated on the prediction of severe hail for a large set of storm patches extracted from the NCAR convection-allowing ensemble. Each storm patch contains information about storm structure and the near-storm environment. Logistic regression and random forest models were trained using the PCA and GAN encodings of the storm data and were compared against the predictions from a convolutional neural network. All methods showed skill over climatology at predicting the probability of severe hail. However, the verification scores among the methods were very similar and the predictions were highly correlated. Further evaluations are being performed to determine how the choice of input variables affects the results.

Stability of deep features across CT scanners and field of view using a physical phantom

NASA Astrophysics Data System (ADS)

Paul, Rahul; Shafiq-ul-Hassan, Muhammad; Moros, Eduardo G.; Gillies, Robert J.; Hall, Lawrence O.; Goldgof, Dmitry B.

2018-02-01

Radiomics is the process of analyzing radiological images by extracting quantitative features for monitoring and diagnosis of various cancers. Analyzing images acquired from different medical centers is confounded by many choices in acquisition, reconstruction parameters and differences among device manufacturers. Consequently, scanning the same patient or phantom using various acquisition/reconstruction parameters as well as different scanners may result in different feature values. To further evaluate this issue, in this study, CT images from a physical radiomic phantom were used. Recent studies showed that some quantitative features were dependent on voxel size and that this dependency could be reduced or removed by the appropriate normalization factor. Deep features extracted from a convolutional neural network, may also provide additional features for image analysis. Using a transfer learning approach, we obtained deep features from three convolutional neural networks pre-trained on color camera images. An we examination of the dependency of deep features on image pixel size was done. We found that some deep features were pixel size dependent, and to remove this dependency we proposed two effective normalization approaches. For analyzing the effects of normalization, a threshold has been used based on the calculated standard deviation and average distance from a best fit horizontal line among the features' underlying pixel size before and after normalization. The inter and intra scanner dependency of deep features has also been evaluated.

Deforestation and Industrial Forest Patterns in Colombia: a Case Study

NASA Astrophysics Data System (ADS)

Huo, L. Z.; Boschetti, L.; Sparks, A. M.; Clerici, N.

2017-12-01

The recent peace agreement between the government and the Revolutionary Armed Forces of Colombia (FARC) offers new opportunities for peaceful and sustainable development, but at the same time requires a timely effort to protect biological resources, and ecosystem services (Clerici et al., 2016). In this context, we use the 2001-2017 Landsat data record to prototype a methodology to establish a baseline of deforestation, afforestation and industrial forest practices (i.e. establishment and harvest of forest plantations), and to monitor future changes. Two study areas, which have seen considerable deforestation in recent years, were selected: one in the South of the country, at the edge of the Amazon Forest (WRS path 008 row 059) and one in the center, in mixed forest (WRS path 008 row 055). The time series of all the available cloud free Landsat 5, Landsat 7 and Landsat 8 data was classified into a sequence of binary forest/non forest maps using a deep learning model, successfully used in the natural language processing field, trained to detect forest transitions. Recurrent Neural Networks (RNN) is a class of artificial neural network that extends the conventional neural network with loops in the connections (Graves et al., 2013). Unlike a feed-forward neural network, an RNN is able to process the sequential inputs by having a recurrent hidden state whose activation at each step depends on that of the previous steps. In this manner, the RNN provides a good framework to dynamically model time series data, and has been successfully applied to natural language processing in Google (Sutskever et al., 2014). The sequence of forest cover state maps was subsequently post-processed to differentiate between deforestation (e.g. transition from forest to non forest land use) and industrial forest harvest (i.e. timber harvest followed by regrowth), by integrating the detection of temporal patterns, and spatial patterns. References Clerici, N., et al., (2016). Colombia: Dealing in conservation. Science, 354(6309), 190-190. Sutskever I.,et al. (2014). Sequence to sequence learning with neural networks. Advances in neural information processing systems, 3104-3112. Graves A., et al. (2013). Speech recognition with deep recurrent neural networks. In Proc. IEEE Int. Conf. Acoust. Speech Signal Process., 6645-6649.

A convolutional neural network to filter artifacts in spectroscopic MRI.

PubMed

Gurbani, Saumya S; Schreibmann, Eduard; Maudsley, Andrew A; Cordova, James Scott; Soher, Brian J; Poptani, Harish; Verma, Gaurav; Barker, Peter B; Shim, Hyunsuk; Cooper, Lee A D

2018-03-09

Proton MRSI is a noninvasive modality capable of generating volumetric maps of in vivo tissue metabolism without the need for ionizing radiation or injected contrast agent. Magnetic resonance spectroscopic imaging has been shown to be a viable imaging modality for studying several neuropathologies. However, a key hurdle in the routine clinical adoption of MRSI is the presence of spectral artifacts that can arise from a number of sources, possibly leading to false information. A deep learning model was developed that was capable of identifying and filtering out poor quality spectra. The core of the model used a tiled convolutional neural network that analyzed frequency-domain spectra to detect artifacts. When compared with a panel of MRS experts, our convolutional neural network achieved high sensitivity and specificity with an area under the curve of 0.95. A visualization scheme was implemented to better understand how the convolutional neural network made its judgement on single-voxel or multivoxel MRSI, and the convolutional neural network was embedded into a pipeline capable of producing whole-brain spectroscopic MRI volumes in real time. The fully automated method for assessment of spectral quality provides a valuable tool to support clinical MRSI or spectroscopic MRI studies for use in fields such as adaptive radiation therapy planning. © 2018 International Society for Magnetic Resonance in Medicine.

Detection of eardrum abnormalities using ensemble deep learning approaches

NASA Astrophysics Data System (ADS)

Senaras, Caglar; Moberly, Aaron C.; Teknos, Theodoros; Essig, Garth; Elmaraghy, Charles; Taj-Schaal, Nazhat; Yua, Lianbo; Gurcan, Metin N.

2018-02-01

In this study, we proposed an approach to report the condition of the eardrum as "normal" or "abnormal" by ensembling two different deep learning architectures. In the first network (Network 1), we applied transfer learning to the Inception V3 network by using 409 labeled samples. As a second network (Network 2), we designed a convolutional neural network to take advantage of auto-encoders by using additional 673 unlabeled eardrum samples. The individual classification accuracies of the Network 1 and Network 2 were calculated as 84.4%(+/- 12.1%) and 82.6% (+/- 11.3%), respectively. Only 32% of the errors of the two networks were the same, making it possible to combine two approaches to achieve better classification accuracy. The proposed ensemble method allows us to achieve robust classification because it has high accuracy (84.4%) with the lowest standard deviation (+/- 10.3%).

End-to-End ASR-Free Keyword Search From Speech

NASA Astrophysics Data System (ADS)

Audhkhasi, Kartik; Rosenberg, Andrew; Sethy, Abhinav; Ramabhadran, Bhuvana; Kingsbury, Brian

2017-12-01

End-to-end (E2E) systems have achieved competitive results compared to conventional hybrid hidden Markov model (HMM)-deep neural network based automatic speech recognition (ASR) systems. Such E2E systems are attractive due to the lack of dependence on alignments between input acoustic and output grapheme or HMM state sequence during training. This paper explores the design of an ASR-free end-to-end system for text query-based keyword search (KWS) from speech trained with minimal supervision. Our E2E KWS system consists of three sub-systems. The first sub-system is a recurrent neural network (RNN)-based acoustic auto-encoder trained to reconstruct the audio through a finite-dimensional representation. The second sub-system is a character-level RNN language model using embeddings learned from a convolutional neural network. Since the acoustic and text query embeddings occupy different representation spaces, they are input to a third feed-forward neural network that predicts whether the query occurs in the acoustic utterance or not. This E2E ASR-free KWS system performs respectably despite lacking a conventional ASR system and trains much faster.

Task-specific feature extraction and classification of fMRI volumes using a deep neural network initialized with a deep belief network: Evaluation using sensorimotor tasks

PubMed Central

Jang, Hojin; Plis, Sergey M.; Calhoun, Vince D.; Lee, Jong-Hwan

2016-01-01

Feedforward deep neural networks (DNN), artificial neural networks with multiple hidden layers, have recently demonstrated a record-breaking performance in multiple areas of applications in computer vision and speech processing. Following the success, DNNs have been applied to neuroimaging modalities including functional/structural magnetic resonance imaging (MRI) and positron-emission tomography data. However, no study has explicitly applied DNNs to 3D whole-brain fMRI volumes and thereby extracted hidden volumetric representations of fMRI that are discriminative for a task performed as the fMRI volume was acquired. Our study applied fully connected feedforward DNN to fMRI volumes collected in four sensorimotor tasks (i.e., left-hand clenching, right-hand clenching, auditory attention, and visual stimulus) undertaken by 12 healthy participants. Using a leave-one-subject-out cross-validation scheme, a restricted Boltzmann machine-based deep belief network was pretrained and used to initialize weights of the DNN. The pretrained DNN was fine-tuned while systematically controlling weight-sparsity levels across hidden layers. Optimal weight-sparsity levels were determined from a minimum validation error rate of fMRI volume classification. Minimum error rates (mean ± standard deviation; %) of 6.9 (± 3.8) were obtained from the three-layer DNN with the sparsest condition of weights across the three hidden layers. These error rates were even lower than the error rates from the single-layer network (9.4 ± 4.6) and the two-layer network (7.4 ± 4.1). The estimated DNN weights showed spatial patterns that are remarkably task-specific, particularly in the higher layers. The output values of the third hidden layer represented distinct patterns/codes of the 3D whole-brain fMRI volume and encoded the information of the tasks as evaluated from representational similarity analysis. Our reported findings show the ability of the DNN to classify a single fMRI volume based on the extraction of hidden representations of fMRI volumes associated with tasks across multiple hidden layers. Our study may be beneficial to the automatic classification/diagnosis of neuropsychiatric and neurological diseases and prediction of disease severity and recovery in (pre-) clinical settings using fMRI volumes without requiring an estimation of activation patterns or ad hoc statistical evaluation. PMID:27079534

Task-specific feature extraction and classification of fMRI volumes using a deep neural network initialized with a deep belief network: Evaluation using sensorimotor tasks.

PubMed

Jang, Hojin; Plis, Sergey M; Calhoun, Vince D; Lee, Jong-Hwan

2017-01-15

Feedforward deep neural networks (DNNs), artificial neural networks with multiple hidden layers, have recently demonstrated a record-breaking performance in multiple areas of applications in computer vision and speech processing. Following the success, DNNs have been applied to neuroimaging modalities including functional/structural magnetic resonance imaging (MRI) and positron-emission tomography data. However, no study has explicitly applied DNNs to 3D whole-brain fMRI volumes and thereby extracted hidden volumetric representations of fMRI that are discriminative for a task performed as the fMRI volume was acquired. Our study applied fully connected feedforward DNN to fMRI volumes collected in four sensorimotor tasks (i.e., left-hand clenching, right-hand clenching, auditory attention, and visual stimulus) undertaken by 12 healthy participants. Using a leave-one-subject-out cross-validation scheme, a restricted Boltzmann machine-based deep belief network was pretrained and used to initialize weights of the DNN. The pretrained DNN was fine-tuned while systematically controlling weight-sparsity levels across hidden layers. Optimal weight-sparsity levels were determined from a minimum validation error rate of fMRI volume classification. Minimum error rates (mean±standard deviation; %) of 6.9 (±3.8) were obtained from the three-layer DNN with the sparsest condition of weights across the three hidden layers. These error rates were even lower than the error rates from the single-layer network (9.4±4.6) and the two-layer network (7.4±4.1). The estimated DNN weights showed spatial patterns that are remarkably task-specific, particularly in the higher layers. The output values of the third hidden layer represented distinct patterns/codes of the 3D whole-brain fMRI volume and encoded the information of the tasks as evaluated from representational similarity analysis. Our reported findings show the ability of the DNN to classify a single fMRI volume based on the extraction of hidden representations of fMRI volumes associated with tasks across multiple hidden layers. Our study may be beneficial to the automatic classification/diagnosis of neuropsychiatric and neurological diseases and prediction of disease severity and recovery in (pre-) clinical settings using fMRI volumes without requiring an estimation of activation patterns or ad hoc statistical evaluation. Copyright © 2016 Elsevier Inc. All rights reserved.

Binary patterns encoded convolutional neural networks for texture recognition and remote sensing scene classification

NASA Astrophysics Data System (ADS)

Anwer, Rao Muhammad; Khan, Fahad Shahbaz; van de Weijer, Joost; Molinier, Matthieu; Laaksonen, Jorma

2018-04-01

Designing discriminative powerful texture features robust to realistic imaging conditions is a challenging computer vision problem with many applications, including material recognition and analysis of satellite or aerial imagery. In the past, most texture description approaches were based on dense orderless statistical distribution of local features. However, most recent approaches to texture recognition and remote sensing scene classification are based on Convolutional Neural Networks (CNNs). The de facto practice when learning these CNN models is to use RGB patches as input with training performed on large amounts of labeled data (ImageNet). In this paper, we show that Local Binary Patterns (LBP) encoded CNN models, codenamed TEX-Nets, trained using mapped coded images with explicit LBP based texture information provide complementary information to the standard RGB deep models. Additionally, two deep architectures, namely early and late fusion, are investigated to combine the texture and color information. To the best of our knowledge, we are the first to investigate Binary Patterns encoded CNNs and different deep network fusion architectures for texture recognition and remote sensing scene classification. We perform comprehensive experiments on four texture recognition datasets and four remote sensing scene classification benchmarks: UC-Merced with 21 scene categories, WHU-RS19 with 19 scene classes, RSSCN7 with 7 categories and the recently introduced large scale aerial image dataset (AID) with 30 aerial scene types. We demonstrate that TEX-Nets provide complementary information to standard RGB deep model of the same network architecture. Our late fusion TEX-Net architecture always improves the overall performance compared to the standard RGB network on both recognition problems. Furthermore, our final combination leads to consistent improvement over the state-of-the-art for remote sensing scene classification.

An Efficient Implementation of Deep Convolutional Neural Networks for MRI Segmentation.

PubMed

Hoseini, Farnaz; Shahbahrami, Asadollah; Bayat, Peyman

2018-02-27

Image segmentation is one of the most common steps in digital image processing, classifying a digital image into different segments. The main goal of this paper is to segment brain tumors in magnetic resonance images (MRI) using deep learning. Tumors having different shapes, sizes, brightness and textures can appear anywhere in the brain. These complexities are the reasons to choose a high-capacity Deep Convolutional Neural Network (DCNN) containing more than one layer. The proposed DCNN contains two parts: architecture and learning algorithms. The architecture and the learning algorithms are used to design a network model and to optimize parameters for the network training phase, respectively. The architecture contains five convolutional layers, all using 3 × 3 kernels, and one fully connected layer. Due to the advantage of using small kernels with fold, it allows making the effect of larger kernels with smaller number of parameters and fewer computations. Using the Dice Similarity Coefficient metric, we report accuracy results on the BRATS 2016, brain tumor segmentation challenge dataset, for the complete, core, and enhancing regions as 0.90, 0.85, and 0.84 respectively. The learning algorithm includes the task-level parallelism. All the pixels of an MR image are classified using a patch-based approach for segmentation. We attain a good performance and the experimental results show that the proposed DCNN increases the segmentation accuracy compared to previous techniques.

DeepQA: improving the estimation of single protein model quality with deep belief networks.

PubMed

Cao, Renzhi; Bhattacharya, Debswapna; Hou, Jie; Cheng, Jianlin

2016-12-05

Protein quality assessment (QA) useful for ranking and selecting protein models has long been viewed as one of the major challenges for protein tertiary structure prediction. Especially, estimating the quality of a single protein model, which is important for selecting a few good models out of a large model pool consisting of mostly low-quality models, is still a largely unsolved problem. We introduce a novel single-model quality assessment method DeepQA based on deep belief network that utilizes a number of selected features describing the quality of a model from different perspectives, such as energy, physio-chemical characteristics, and structural information. The deep belief network is trained on several large datasets consisting of models from the Critical Assessment of Protein Structure Prediction (CASP) experiments, several publicly available datasets, and models generated by our in-house ab initio method. Our experiments demonstrate that deep belief network has better performance compared to Support Vector Machines and Neural Networks on the protein model quality assessment problem, and our method DeepQA achieves the state-of-the-art performance on CASP11 dataset. It also outperformed two well-established methods in selecting good outlier models from a large set of models of mostly low quality generated by ab initio modeling methods. DeepQA is a useful deep learning tool for protein single model quality assessment and protein structure prediction. The source code, executable, document and training/test datasets of DeepQA for Linux is freely available to non-commercial users at http://cactus.rnet.missouri.edu/DeepQA/ .

Simulation of noisy dynamical system by Deep Learning

NASA Astrophysics Data System (ADS)

Yeo, Kyongmin

2017-11-01

Deep learning has attracted huge attention due to its powerful representation capability. However, most of the studies on deep learning have been focused on visual analytics or language modeling and the capability of the deep learning in modeling dynamical systems is not well understood. In this study, we use a recurrent neural network to model noisy nonlinear dynamical systems. In particular, we use a long short-term memory (LSTM) network, which constructs internal nonlinear dynamics systems. We propose a cross-entropy loss with spatial ridge regularization to learn a non-stationary conditional probability distribution from a noisy nonlinear dynamical system. A Monte Carlo procedure to perform time-marching simulations by using the LSTM is presented. The behavior of the LSTM is studied by using noisy, forced Van der Pol oscillator and Ikeda equation.

Modelling Peri-Perceptual Brain Processes in a Deep Learning Spiking Neural Network Architecture.

PubMed

Gholami Doborjeh, Zohreh; Kasabov, Nikola; Gholami Doborjeh, Maryam; Sumich, Alexander

2018-06-11

Familiarity of marketing stimuli may affect consumer behaviour at a peri-perceptual processing level. The current study introduces a method for deep learning of electroencephalogram (EEG) data using a spiking neural network (SNN) approach that reveals the complexity of peri-perceptual processes of familiarity. The method is applied to data from 20 participants viewing familiar and unfamiliar logos. The results support the potential of SNN models as novel tools in the exploration of peri-perceptual mechanisms that respond differentially to familiar and unfamiliar stimuli. Specifically, the activation pattern of the time-locked response identified by the proposed SNN model at approximately 200 milliseconds post-stimulus suggests greater connectivity and more widespread dynamic spatio-temporal patterns for familiar than unfamiliar logos. The proposed SNN approach can be applied to study other peri-perceptual or perceptual brain processes in cognitive and computational neuroscience.

a Novel Deep Convolutional Neural Network for Spectral-Spatial Classification of Hyperspectral Data

NASA Astrophysics Data System (ADS)

Li, N.; Wang, C.; Zhao, H.; Gong, X.; Wang, D.

2018-04-01

Spatial and spectral information are obtained simultaneously by hyperspectral remote sensing. Joint extraction of these information of hyperspectral image is one of most import methods for hyperspectral image classification. In this paper, a novel deep convolutional neural network (CNN) is proposed, which extracts spectral-spatial information of hyperspectral images correctly. The proposed model not only learns sufficient knowledge from the limited number of samples, but also has powerful generalization ability. The proposed framework based on three-dimensional convolution can extract spectral-spatial features of labeled samples effectively. Though CNN has shown its robustness to distortion, it cannot extract features of different scales through the traditional pooling layer that only have one size of pooling window. Hence, spatial pyramid pooling (SPP) is introduced into three-dimensional local convolutional filters for hyperspectral classification. Experimental results with a widely used hyperspectral remote sensing dataset show that the proposed model provides competitive performance.

Beam-column joint shear prediction using hybridized deep learning neural network with genetic algorithm

NASA Astrophysics Data System (ADS)

Mundher Yaseen, Zaher; Abdulmohsin Afan, Haitham; Tran, Minh-Tung

2018-04-01

Scientifically evidenced that beam-column joints are a critical point in the reinforced concrete (RC) structure under the fluctuation loads effects. In this novel hybrid data-intelligence model developed to predict the joint shear behavior of exterior beam-column structure frame. The hybrid data-intelligence model is called genetic algorithm integrated with deep learning neural network model (GA-DLNN). The genetic algorithm is used as prior modelling phase for the input approximation whereas the DLNN predictive model is used for the prediction phase. To demonstrate this structural problem, experimental data is collected from the literature that defined the dimensional and specimens’ properties. The attained findings evidenced the efficitveness of the hybrid GA-DLNN in modelling beam-column joint shear problem. In addition, the accurate prediction achived with less input variables owing to the feasibility of the evolutionary phase.

Enhancing deep convolutional neural network scheme for breast cancer diagnosis with unlabeled data.

PubMed

Sun, Wenqing; Tseng, Tzu-Liang Bill; Zhang, Jianying; Qian, Wei

2017-04-01

In this study we developed a graph based semi-supervised learning (SSL) scheme using deep convolutional neural network (CNN) for breast cancer diagnosis. CNN usually needs a large amount of labeled data for training and fine tuning the parameters, and our proposed scheme only requires a small portion of labeled data in training set. Four modules were included in the diagnosis system: data weighing, feature selection, dividing co-training data labeling, and CNN. 3158 region of interests (ROIs) with each containing a mass extracted from 1874 pairs of mammogram images were used for this study. Among them 100 ROIs were treated as labeled data while the rest were treated as unlabeled. The area under the curve (AUC) observed in our study was 0.8818, and the accuracy of CNN is 0.8243 using the mixed labeled and unlabeled data. Copyright © 2016. Published by Elsevier Ltd.

Enhancing Hi-C data resolution with deep convolutional neural network HiCPlus.

PubMed

Zhang, Yan; An, Lin; Xu, Jie; Zhang, Bo; Zheng, W Jim; Hu, Ming; Tang, Jijun; Yue, Feng

2018-02-21

Although Hi-C technology is one of the most popular tools for studying 3D genome organization, due to sequencing cost, the resolution of most Hi-C datasets are coarse and cannot be used to link distal regulatory elements to their target genes. Here we develop HiCPlus, a computational approach based on deep convolutional neural network, to infer high-resolution Hi-C interaction matrices from low-resolution Hi-C data. We demonstrate that HiCPlus can impute interaction matrices highly similar to the original ones, while only using 1/16 of the original sequencing reads. We show that the models learned from one cell type can be applied to make predictions in other cell or tissue types. Our work not only provides a computational framework to enhance Hi-C data resolution but also reveals features underlying the formation of 3D chromatin interactions.

Image annotation by deep neural networks with attention shaping

NASA Astrophysics Data System (ADS)

Zheng, Kexin; Lv, Shaohe; Ma, Fang; Chen, Fei; Jin, Chi; Dou, Yong

2017-07-01

Image annotation is a task of assigning semantic labels to an image. Recently, deep neural networks with visual attention have been utilized successfully in many computer vision tasks. In this paper, we show that conventional attention mechanism is easily misled by the salient class, i.e., the attended region always contains part of the image area describing the content of salient class at different attention iterations. To this end, we propose a novel attention shaping mechanism, which aims to maximize the non-overlapping area between consecutive attention processes by taking into account the history of previous attention vectors. Several weighting polices are studied to utilize the history information in different manners. In two benchmark datasets, i.e., PASCAL VOC2012 and MIRFlickr-25k, the average precision is improved by up to 10% in comparison with the state-of-the-art annotation methods.

An Adaptive Multi-Sensor Data Fusion Method Based on Deep Convolutional Neural Networks for Fault Diagnosis of Planetary Gearbox

PubMed Central

Jing, Luyang; Wang, Taiyong; Zhao, Ming; Wang, Peng

2017-01-01

A fault diagnosis approach based on multi-sensor data fusion is a promising tool to deal with complicated damage detection problems of mechanical systems. Nevertheless, this approach suffers from two challenges, which are (1) the feature extraction from various types of sensory data and (2) the selection of a suitable fusion level. It is usually difficult to choose an optimal feature or fusion level for a specific fault diagnosis task, and extensive domain expertise and human labor are also highly required during these selections. To address these two challenges, we propose an adaptive multi-sensor data fusion method based on deep convolutional neural networks (DCNN) for fault diagnosis. The proposed method can learn features from raw data and optimize a combination of different fusion levels adaptively to satisfy the requirements of any fault diagnosis task. The proposed method is tested through a planetary gearbox test rig. Handcraft features, manual-selected fusion levels, single sensory data, and two traditional intelligent models, back-propagation neural networks (BPNN) and a support vector machine (SVM), are used as comparisons in the experiment. The results demonstrate that the proposed method is able to detect the conditions of the planetary gearbox effectively with the best diagnosis accuracy among all comparative methods in the experiment. PMID:28230767

Using Deep Learning to Analyze the Voices of Stars.

NASA Astrophysics Data System (ADS)

Boudreaux, Thomas Macaulay

2018-01-01

With several new large-scale surveys on the horizon, including LSST, TESS, ZTF, and Evryscope, faster and more accurate analysis methods will be required to adequately process the enormous amount of data produced. Deep learning, used in industry for years now, allows for advanced feature detection in minimally prepared datasets at very high speeds; however, despite the advantages of this method, its application to astrophysics has not yet been extensively explored. This dearth may be due to a lack of training data available to researchers. Here we generate synthetic data loosely mimicking the properties of acoustic mode pulsating stars and compare the performance of different deep learning algorithms, including Artifical Neural Netoworks, and Convolutional Neural Networks, in classifing these synthetic data sets as either pulsators, or not observed to vary stars.

Deep convolutional neural network for classifying Fusarium wilt of radish from unmanned aerial vehicles

NASA Astrophysics Data System (ADS)

Ha, Jin Gwan; Moon, Hyeonjoon; Kwak, Jin Tae; Hassan, Syed Ibrahim; Dang, Minh; Lee, O. New; Park, Han Yong

2017-10-01

Recently, unmanned aerial vehicles (UAVs) have gained much attention. In particular, there is a growing interest in utilizing UAVs for agricultural applications such as crop monitoring and management. We propose a computerized system that is capable of detecting Fusarium wilt of radish with high accuracy. The system adopts computer vision and machine learning techniques, including deep learning, to process the images captured by UAVs at low altitudes and to identify the infected radish. The whole radish field is first segmented into three distinctive regions (radish, bare ground, and mulching film) via a softmax classifier and K-means clustering. Then, the identified radish regions are further classified into healthy radish and Fusarium wilt of radish using a deep convolutional neural network (CNN). In identifying radish, bare ground, and mulching film from a radish field, we achieved an accuracy of ≥97.4%. In detecting Fusarium wilt of radish, the CNN obtained an accuracy of 93.3%. It also outperformed the standard machine learning algorithm, obtaining 82.9% accuracy. Therefore, UAVs equipped with computational techniques are promising tools for improving the quality and efficiency of agriculture today.

Named Entity Recognition in Chinese Clinical Text Using Deep Neural Network.

PubMed

Wu, Yonghui; Jiang, Min; Lei, Jianbo; Xu, Hua

2015-01-01

Rapid growth in electronic health records (EHRs) use has led to an unprecedented expansion of available clinical data in electronic formats. However, much of the important healthcare information is locked in the narrative documents. Therefore Natural Language Processing (NLP) technologies, e.g., Named Entity Recognition that identifies boundaries and types of entities, has been extensively studied to unlock important clinical information in free text. In this study, we investigated a novel deep learning method to recognize clinical entities in Chinese clinical documents using the minimal feature engineering approach. We developed a deep neural network (DNN) to generate word embeddings from a large unlabeled corpus through unsupervised learning and another DNN for the NER task. The experiment results showed that the DNN with word embeddings trained from the large unlabeled corpus outperformed the state-of-the-art CRF's model in the minimal feature engineering setting, achieving the highest F1-score of 0.9280. Further analysis showed that word embeddings derived through unsupervised learning from large unlabeled corpus remarkably improved the DNN with randomized embedding, denoting the usefulness of unsupervised feature learning.

Deep hierarchical attention network for video description

NASA Astrophysics Data System (ADS)

Li, Shuohao; Tang, Min; Zhang, Jun

2018-03-01

Pairing video to natural language description remains a challenge in computer vision and machine translation. Inspired by image description, which uses an encoder-decoder model for reducing visual scene into a single sentence, we propose a deep hierarchical attention network for video description. The proposed model uses convolutional neural network (CNN) and bidirectional LSTM network as encoders while a hierarchical attention network is used as the decoder. Compared to encoder-decoder models used in video description, the bidirectional LSTM network can capture the temporal structure among video frames. Moreover, the hierarchical attention network has an advantage over single-layer attention network on global context modeling. To make a fair comparison with other methods, we evaluate the proposed architecture with different types of CNN structures and decoders. Experimental results on the standard datasets show that our model has a more superior performance than the state-of-the-art techniques.

Hierarchical modeling of molecular energies using a deep neural network

NASA Astrophysics Data System (ADS)

Lubbers, Nicholas; Smith, Justin S.; Barros, Kipton

2018-06-01

We introduce the Hierarchically Interacting Particle Neural Network (HIP-NN) to model molecular properties from datasets of quantum calculations. Inspired by a many-body expansion, HIP-NN decomposes properties, such as energy, as a sum over hierarchical terms. These terms are generated from a neural network—a composition of many nonlinear transformations—acting on a representation of the molecule. HIP-NN achieves the state-of-the-art performance on a dataset of 131k ground state organic molecules and predicts energies with 0.26 kcal/mol mean absolute error. With minimal tuning, our model is also competitive on a dataset of molecular dynamics trajectories. In addition to enabling accurate energy predictions, the hierarchical structure of HIP-NN helps to identify regions of model uncertainty.

Deep neural nets as a method for quantitative structure-activity relationships.

PubMed

Ma, Junshui; Sheridan, Robert P; Liaw, Andy; Dahl, George E; Svetnik, Vladimir

2015-02-23

Neural networks were widely used for quantitative structure-activity relationships (QSAR) in the 1990s. Because of various practical issues (e.g., slow on large problems, difficult to train, prone to overfitting, etc.), they were superseded by more robust methods like support vector machine (SVM) and random forest (RF), which arose in the early 2000s. The last 10 years has witnessed a revival of neural networks in the machine learning community thanks to new methods for preventing overfitting, more efficient training algorithms, and advancements in computer hardware. In particular, deep neural nets (DNNs), i.e. neural nets with more than one hidden layer, have found great successes in many applications, such as computer vision and natural language processing. Here we show that DNNs can routinely make better prospective predictions than RF on a set of large diverse QSAR data sets that are taken from Merck's drug discovery effort. The number of adjustable parameters needed for DNNs is fairly large, but our results show that it is not necessary to optimize them for individual data sets, and a single set of recommended parameters can achieve better performance than RF for most of the data sets we studied. The usefulness of the parameters is demonstrated on additional data sets not used in the calibration. Although training DNNs is still computationally intensive, using graphical processing units (GPUs) can make this issue manageable.

Detection and diagnosis of colitis on computed tomography using deep convolutional neural networks.

PubMed

Liu, Jiamin; Wang, David; Lu, Le; Wei, Zhuoshi; Kim, Lauren; Turkbey, Evrim B; Sahiner, Berkman; Petrick, Nicholas A; Summers, Ronald M

2017-09-01

Colitis refers to inflammation of the inner lining of the colon that is frequently associated with infection and allergic reactions. In this paper, we propose deep convolutional neural networks methods for lesion-level colitis detection and a support vector machine (SVM) classifier for patient-level colitis diagnosis on routine abdominal CT scans. The recently developed Faster Region-based Convolutional Neural Network (Faster RCNN) is utilized for lesion-level colitis detection. For each 2D slice, rectangular region proposals are generated by region proposal networks (RPN). Then, each region proposal is jointly classified and refined by a softmax classifier and bounding-box regressor. Two convolutional neural networks, eight layers of ZF net and 16 layers of VGG net are compared for colitis detection. Finally, for each patient, the detections on all 2D slices are collected and a SVM classifier is applied to develop a patient-level diagnosis. We trained and evaluated our method with 80 colitis patients and 80 normal cases using 4 × 4-fold cross validation. For lesion-level colitis detection, with ZF net, the mean of average precisions (mAP) were 48.7% and 50.9% for RCNN and Faster RCNN, respectively. The detection system achieved sensitivities of 51.4% and 54.0% at two false positives per patient for RCNN and Faster RCNN, respectively. With VGG net, Faster RCNN increased the mAP to 56.9% and increased the sensitivity to 58.4% at two false positive per patient. For patient-level colitis diagnosis, with ZF net, the average areas under the ROC curve (AUC) were 0.978 ± 0.009 and 0.984 ± 0.008 for RCNN and Faster RCNN method, respectively. The difference was not statistically significant with P = 0.18. At the optimal operating point, the RCNN method correctly identified 90.4% (72.3/80) of the colitis patients and 94.0% (75.2/80) of normal cases. The sensitivity improved to 91.6% (73.3/80) and the specificity improved to 95.0% (76.0/80) for the Faster RCNN method. With VGG net, Faster RCNN increased the AUC to 0.986 ± 0.007 and increased the diagnosis sensitivity to 93.7% (75.0/80) and specificity was unchanged at 95.0% (76.0/80). Colitis detection and diagnosis by deep convolutional neural networks is accurate and promising for future clinical application. Published 2017. This article is a U.S. Government work and is in the public domain in the USA.