Jankovic, Marko V; Ogawa, Hidemitsu
This paper presents analysis of the recently proposed modulated Hebb-Oja (MHO) method that performs linear mapping to a lower-dimensional subspace. Principal component subspace is the method that will be analyzed. Comparing to some other well-known methods for yielding principal component subspace (e.g., Oja's Subspace Learning Algorithm), the proposed method has one feature that could be seen as desirable from the biological point of view--synaptic efficacy learning rule does not need the explicit information about the value of the other efficacies to make individual efficacy modification. Also, the simplicity of the "neural circuits" that perform global computations and a fact that their number does not depend on the number of input and output neurons, could be seen as good features of the proposed method. PMID:16566463
Walz, Garry R., Ed.; Knowdell, Richard L., Ed.
This publication contains papers based on program presentations from the 2003 International Career Development Conference. Chapters include: (1) How to Turn Your Passion into a Profit (S. Abbott); (2) Harnessing the Power of Career Transition Groups (M. Adoradio and A. Oja); (3) All the Worlds a Stage Using Theatre in Career Counseling (P.…
Yoder, Paul; Woynaroski, Tiffany; Fey, Marc; Warren, Steven
Children with intellectual disability were randomly assigned to receive Milieu Communication Teaching (MCT) at one 1-hr session per week (low dose frequency, LDF) or five 1-hr sessions per week (high dose frequency, HDF) over 9 months (Fey, Yoder, Warren, & Bredin-Oja, 2013. Non-Down syndrome (NDS) and Down syndrome (DS) subgroups were matched…
When the independent sources are known to be nonnegative and well-grounded, which means that they have a nonzero pdf in the region of zero, Oja and Plumbley have proposed a "Nonnegative principal component analysis (PCA)" algorithm to separate these positive sources. Generally, it is very difficult to prove the convergence of a discrete-time independent component analysis (ICA) learning algorithm. However, by using the skew-symmetry property of this discrete-time "Nonnegative PCA" algorithm, if the learning rate satisfies suitable condition, the global convergence of this discrete-time algorithm can be proven. Simulation results are employed to further illustrate the advantages of this theory. PMID:16526495
Han, Fang; Wiercigroch, Marian; Fang, Jian-An; Wang, Zhijie
Excitement and synchronization of electrically and chemically coupled Newman-Watts (NW) small-world neuronal networks with a short-term synaptic plasticity described by a modified Oja learning rule are investigated. For each type of neuronal network, the variation properties of synaptic weights are examined first. Then the effects of the learning rate, the coupling strength and the shortcut-adding probability on excitement and synchronization of the neuronal network are studied. It is shown that the synaptic learning suppresses the over-excitement, helps synchronization for the electrically coupled network but impairs synchronization for the chemically coupled one. Both the introduction of shortcuts and the increase of the coupling strength improve synchronization and they are helpful in increasing the excitement for the chemically coupled network, but have little effect on the excitement of the electrically coupled one. PMID:21956933
Yoder, Paul; Woynaroski, Tiffany; Fey, Marc; Warren, Steven
Children with intellectual disability were randomly assigned to receive Milieu Communication Teaching (MCT) at one 1-hr session per week (low dose frequency, LDF) or five 1-hr sessions per week (high dose frequency, HDF) over 9 months ( Fey, Yoder, Warren, & Bredin-Oja, 2013 . Non-Down syndrome (NDS) and Down syndrome (DS) subgroups were matched on intelligence, mental age, and chronological age. The NDS group had significantly more growth in spoken vocabulary than the DS group. In the DS subgroup, the HDF group had more spoken vocabulary growth than the LDF group when IQ was controlled. In both etiological subgroups, the HDF group yielded greater vocabulary production outcomes than the LDF group for children who played functionally with a range of objects. PMID:24450319
Rao, A. Venkoba
Ayurveda now among the alternative complementary systems of medicine is over 5000 years old. Its origin and the compilation of Caraka Samhita are noted. The nature of mind as a sensory and a motor organ, its structure and functions are discussed. The concept of Thdosha theory and Trigunas are explained besides the so-called master-forms of Doshas namely Prana, Tejas and Ojas. The constituional and tempermental types depending upon the doshas are described. These determine diagnoses and guide treatment. Ayurveda is highlighted as a holistic system with its concern for prevention of disease and promotion of health. Disease denotes failure of prophylaxis. Some methods of Ayurvedic therapy are mentioned. PMID:21206574
Jankovic, Marko; Ogawa, Hidemitsu
This paper presents one possible implementation of a transformation that performs linear mapping to a lower-dimensional subspace. Principal component subspace will be the one that will be analyzed. Idea implemented in this paper represents generalization of the recently proposed infinity OH neural method for principal component extraction. The calculations in the newly proposed method are performed locally--a feature which is usually considered as desirable from the biological point of view. Comparing to some other wellknown methods, proposed synaptic efficacy learning rule requires less information about the value of the other efficacies to make single efficacy modification. Synaptic efficacies are modified by implementation of Modulated Hebb-type (MH) learning rule. Slightly modified MH algorithm named Modulated Hebb Oja (MHO) algorithm, will be also introduced. Structural similarity of the proposed network with part of the retinal circuit will be presented, too. PMID:12964209
Kumar, Sanjai; Singh, Girish; Pandey, Ajai Kumar; Singh, Ram Harsh
Background: Recent years have shown an alarming rise in the incidence of diabetes mellitus (DM) all over the world. The present management of DM it not satisfactory. Hence, alternative systems of medicine are also being explored. Prameha as described in Ayurveda is a disease synonymous with today's DM. The patients of Prameha inherently carry the risk of impaired Agni and depleted Ojas status, that is, hypometabolic and immuno-compromised state. Now the primary goal is not merely to achieve normoglycemia, but also to minimize its complications. In this context, many Ayurvedic drugs are undergoing extensive research. Aim: To evaluate the anti-diabetic, immune-enhancer and biofire balancing effects of Naimittika Rasayana drugs viz. Silajatu and Mamajjaka in type-2 DM. Materials and Methods: A total of 95 patients of type-2 DM were registered; in which 84 patients turned up for full follow-up. Patients were randomly allocated into three groups; Group-A was treated with Mamajjaka (500mg twice a day) and Group-B with Silajatu (500mg twice a day) and Group-C was treated with modern drug and assessment was done at monthly intervals for three months. Results: The selected Rasayana drugs have shown good response on subjective and objective parameters. The Mamajjaka treated patients responded better. However, as regards the reduction of post prandial blood sugar, Silajatu was superior. Conclusion: The Ayurveda-inspired holistic approach seems to have a unique response promoting Agni (biofire) and Ojas (immune strength) status leading to good health and wellness. PMID:26195903
Adebayo-Oyetoro, A O; Oyewole, O B; Obadina, A O; Omemu, M A
The microorganisms involved in the fermentation and spoilage of fermented cassava flour were investigated. The water samples used at the different processing sites were also investigated to determine their safety status. There was predominance of Staphylococcus aureus, Aspergillus spp., and Escherichia coli in all samples. Coliforms were observed to be present in all of the processing water. In the fermented cassava flour, the total bacterial count ranged between 4.9 × 10(6) cfu/mL from Eleso, Bakatari, and Oja Odan processing sites and 8.10 × 10(6) cfu/mL in Eruku processing site. The majority of the microorganisms involved in the spoilage of "lafun" were found to be Aspergillus niger which ranged between 4.6 × 10(5) cfu/mL in Eleso and 8.1 × 10(5) cfu/mL in Kila. The control sample prepared in the laboratory had a low microbial load compared to samples collected from various sites and markets. PMID:26904609
Adebayo-Oyetoro, A. O.; Oyewole, O. B.; Obadina, A. O.; Omemu, M. A.
The microorganisms involved in the fermentation and spoilage of fermented cassava flour were investigated. The water samples used at the different processing sites were also investigated to determine their safety status. There was predominance of Staphylococcus aureus, Aspergillus spp., and Escherichia coli in all samples. Coliforms were observed to be present in all of the processing water. In the fermented cassava flour, the total bacterial count ranged between 4.9 × 106 cfu/mL from Eleso, Bakatari, and Oja Odan processing sites and 8.10 × 106 cfu/mL in Eruku processing site. The majority of the microorganisms involved in the spoilage of “lafun” were found to be Aspergillus niger which ranged between 4.6 × 105 cfu/mL in Eleso and 8.1 × 105 cfu/mL in Kila. The control sample prepared in the laboratory had a low microbial load compared to samples collected from various sites and markets. PMID:26904609
Yoder, Paul J.; Woynaroski, Tiffany G.; Fey, Marc E.; Warren, Steven F.
This study involves a re-analysis of spoken vocabulary outcomes of children with intellectual disabilities who were randomly assigned to receive Milieu Communication Teaching (MCT) at low (one 1-hour session per week) or high (five 1-hour sessions per week) dose frequency over nine months (Fey, Yoder, Warren, & Bredin-Oja, in press). Non-Down syndrome (NDS) and Down syndrome (DS) subgroups were matched on intelligence, mental age, and chronological age. A growth model including intercept, slope, and quadratic revealed that children in the NDS group had significantly more growth in spoken vocabulary than children in the DS group independent of dose frequency manipulations. Subsequent etiological subgroup analyses demonstrated that in the DS subgroup, children receiving MCT at the higher dose frequency had more spoken vocabulary growth than children receiving MCT at the lower dose frequency. Subgroup analyses also supported our previous findings that high dose frequency of MCT yielded greater vocabulary production outcomes than low dose frequency for children who played functionally with a range of objects, regardless of etiology. PMID:24450319
Lahti, I.; Korja, T.; Petersen, L. B.
LITHOSPHERIC CONDUCTIVITY ALONG THE GGT/SVEKA TRANSECT IN THE FENNOSCANDIAN SHIELD I. Lahti (1), T. Korja (1), L. Pedersen (2) and BEAR Working Group (1) Department of Geophysics, University of Oulu, Finland (2) Department of Earth Sciences, University of Uppsala, Sweden firstname.lastname@example.org The GGT/SVEKA transect traverses the main tectonic units in the central part of the Fennoscandian Shield in NE-SW direction. These units are the Archaean Karelian Province in the northeast and several Palaeoproterozoic arc complexes in the Svecofennian Domain in the southwest. Since 1985 over 150 magnetotelluric (MT) soundings of which 140 are short period and 10 long period soundings have been made in the survey area. We have performed several 2-D Occam inversions of the MT data using the REBOCC code (Siripunvaraporn and Egbert, 2000) to generate smooth conductivity models for the survey area. The best fitting model with the RMS error below 3.0% is obtained by using the determinant of impedance tensor as the inverted parameter. Highly conductive dipping conductors at both sides of the boundary zone between the arc complexes in southern and central Finland are seen in the final model. Both conductors represent borders of major crustal segments possibly indicating two subductions in the research area. In contrast, only minor conductivity variations are seen at the lithological boundary between the Karelian and Svecofennian domains in central Finland whereas a southwestward dipping conductor beneath the Palaeoproterozoic Kainuu Belt is revealed. The conductor suggests the presence of Palaeoproterozoic graphite bearing sedimentary rocks beneath the Archaean rocks of the Iisalmi complex. Lower crustal conductor is absent to NE from the Kainuu Belt while the conductor is present in the Palaeoproterozoic Svecofennian Domain to southwest from the Kainuu Belt. Thus, the main conductivity boundary between the Archaean and Palaeoproterozoic lithosphere is located beneath the Kainuu Belt
Steyn-Ross, D A; Steyn-Ross, M L; Wilcocks, L C; Sleigh, J W
In our two recent papers [M.L. Steyn-Ross et al., Phys. Rev. E 60, 7299 (1999); 64, 011917 (2001)] we presented clinical evidence for a general anesthetic-induced phase change in the cerebral cortex, and showed how the significant features of the cortical phase change (biphasic power surge, spectral energy redistribution, "heat capacity" divergence), could be explained using a stochastic single-macrocolumn model of the cortex. The model predictions were based on rather strong "adiabatic" assumptions which assert that the mean-field excitatory and inhibitory macrocolumn voltages are "slow" variables whose equilibration times are much longer than those of the input "currents" that drive the macrocolumn. In the present paper we test the adiabatic assumption by running numerical simulations of the stochastic differential equations. These simulations confirm the number and nature of the steady-state solutions, the growth of fluctuation power at transition, and the redistribution of spectral energy towards lower frequencies. We use spectral entropy to quantify these changes in the power spectral density, and to show that the spectral entropy should decrease markedly at the point of transition. This prediction agrees with recent clinical findings by Viertiö-Oja and colleagues [J. Clinical Monitoring Computing 16, 60 (2000)]. Our modeling work shows that there is an inverse relationship between spectral entropy H and correlation time T of the soma-voltage fluctuations: H inversely proportional to (ln T). In a theoretical analysis we prove that this proportionality becomes exact for an ideal Lorentzian process. These findings suggest that by monitoring the changes in EEG correlation time, it should be possible to track changes in the state of patient consciousness. PMID:11461299
Laisk, Agu; Oja, Vello; Eichelmann, Hillar; Dall'Osto, Luca
The spectral global quantum yield (YII, electrons/photons absorbed) of photosystem II (PSII) was measured in sunflower leaves in State 1 using monochromatic light. The global quantum yield of PSI (YI) was measured using low-intensity monochromatic light flashes and the associated transmittance change at 810nm. The 810-nm signal change was calibrated based on the number of electrons generated by PSII during the flash (4·O2 evolution) which arrived at the PSI donor side after a delay of 2ms. The intrinsic quantum yield of PSI (yI, electrons per photon absorbed by PSI) was measured at 712nm, where photon absorption by PSII was small. The results were used to resolve the individual spectra of the excitation partitioning coefficients between PSI (aI) and PSII (aII) in leaves. For comparison, pigment-protein complexes for PSII and PSI were isolated, separated by sucrose density ultracentrifugation, and their optical density was measured. A good correlation was obtained for the spectral excitation partitioning coefficients measured by these different methods. The intrinsic yield of PSI was high (yI=0.88), but it absorbed only about 1/3 of quanta; consequently, about 2/3 of quanta were absorbed by PSII, but processed with the low intrinsic yield yII=0.63. In PSII, the quantum yield of charge separation was 0.89 as detected by variable fluorescence Fv/Fm, but 29% of separated charges recombined (Laisk A, Eichelmann H and Oja V, Photosynth. Res. 113, 145-155). At wavelengths less than 580nm about 30% of excitation is absorbed by pigments poorly connected to either photosystem, most likely carotenoids bound in pigment-protein complexes. PMID:24333386
Laubach, M; Shuler, M; Nicolelis, M A
The goal of this study was to compare how multivariate statistical methods for dimension reduction account for correlations between simultaneously recorded neurons. Here, we describe applications of principal component analysis (PCA) and independent component analysis (ICA) (Cardoso J-F, Souloumiac A. IEE-Proc F 1993;140:362-70; Hyvarinen A, Oja E. Neural Comput 1997;9:1483-92; Lee TW, Girolami M, Sejnowski TJ. Neural Comp 1999;11:417-41) to neuronal ensemble data. Simulated ensembles of neurons were used to compare how well the methods above could account for correlated neuronal firing. The simulations showed that 'population vectors' defined by PCA were broadly distributed over the neuronal ensembles; thus, PCA was unable to identify independent groupings of neurons that shared common sources of input. By contrast, the ICA methods were all able to identify groupings of neurons that emerged due to correlated firing. This result suggests that correlated neuronal firing is reflected in higher-order correlations between neurons and not simply in the neurons' covariance. To assess the significance of these methods for real neuronal ensembles, we analyzed data from populations of neurons recorded in the motor cortex of rats trained to perform a reaction-time task. Scores for PCA and ICA were reconstructed on a bin-by-bin basis for single trials. These data were then used to train an artificial neural network to discriminate between single trials with either short or long reaction-times. Classifications based on scores from the ICA-based methods were significantly better than those based on PCA. For example, scores for components defined with an ICA-based method, extended ICA (Lee et al., 1999), classified more trials correctly (80.58+/-1.25%) than PCA (73.14+/-0.84%) for an ensemble of 26 neurons recorded in the motor cortex (ANOVA: P < 0.005). This result suggests that behaviorally relevant information is represented in correlated neuronal firing and can be best
Adeyemi, Adewale Samson; Adebayo, Philip Babatunde; Tanimowo, Moses O.; Ayodele, Olugbenga Edward
Background Allergic disorders have become a major public health concern worldwide. No Nigerian study has examined the epidemiology of allergic diseases among women. Aim To document the prevalence, risk factors and the changes in the symptoms of allergic disorders during pregnancy. Settings and Design Cross-sectional study conducted at the booking and antenatal clinics of LAUTECH Teaching Hospital and Millennium Development Goals (MDG) Clinic of the Comprehensive Health Center, Oja Igbo, Ogbomoso, Nigeria. Materials and Methods Study enrolled 432 women from two public hospitals. Sociodemographic and clinical history were obtained and allergic disorders were diagnosed using ISAAC questionnaires. Results The prevalence of wheezing, eczema and rhinitis in pregnancy are 7.5%, 4.0% and 5.8% respectively. The prevalence of wheezing and eczema was slightly higher among the pregnant in past 12 months. Wheeze worsened in 70% (18/26), improved in 15% (2/26), and stable in 15% (2/26). Eczema worsened in 50% (7/14), improved in 7.1% (1/14) and stable in 42.9% (6/14), while allergic rhinitis worsened in 50% (11/22), improved in 22.7% (5/22) and stabilized in 27.3 % (6/22). In multivariate analysis, the risk of allergic diseases in pregnancy was increase 2 times by low income earning (CI: 1.2 – 2.1, p = 0.002), low level education (OR = 0.6, CI: 0.3 – 0.9, p = 0.011) and by family history of asthma, OR-4.3, CI – 1.3 – 13.9, p = 0.015. Family history of asthma increase the chances of asthma by 18.7 times, CI-2.3 – 152.2, p = 0.006, while the odd of eczema was increased 9.1 times (CI-2.7 – 30.6, p<0.001) and 2.4 times (CI: 1.2 – 4.7, p = 0.008) by second hand home smoking and low-family income respectively. The risk of allergic rhinitis were raised 1.8 times by low family income (CI 1.1 – 2.8, p = 0.013) and 3.9 times by family history of rhinitis (OR = 3.9, CI 1.2 – 12.7, p = 0.024). Conclusion Prevalence of wheezing and eczema are higher in pregnancy probably
The effects of low and high nitrogen diets on amino acid levels were studied in hydrolyzates of ruminal bacteria adhered to four topographically different anatomic parts of the ruminal wall (dorsal, ventral and caudal parts as well as reticulum) in 18 sheep + of the Slovak Merino breed divided into three experimental groups. The epimural bacteria of the dorsal and ventral parts of the ovine rumen revealed the most sensitive reaction to the varying amounts of nitrogen ingested with the diet. In hydrolyzates of ruminal bacteria adhered to the dorsal and ventral epithelium, 15 and 14 amino acids were changing (Figs. 1, 2). In hydrolyzates of epimural bacteria, a sensitive reaction was observed in the following amino acids: alanine, histidine, thyroxin, arginine and proline (Tabs. I-IV). In all topographical and anatomical parts of the rumen, both alanine and histidine levels in hydrolyzates of epimural ruminal bacteria significantly increased with the diet with high-nitrogen content fed, but was falling in sheep fed with low-nitrogen diet. Changes in alanine concentrations may be explained by the fact that alanine forms a part of the mechanism for short-time storage of ammonia in bacterial cells (Bartos, 1987). The fact that alanine is in its lack deaminated to pyruvate (Havassy, 1976) is explained by significant fall in alanine contained in hydrolyzates of bacterial proteins when fed low nitrogen diets. Significant fall in alanine in shortage of amino acid bound nitrogen can be explained by the fact that under these conditions, the alanine skeleton is being incorporated in to 80% of amino acids synthetized de novo by ruminal bacteria (Syväoja and Kreula, 1980). When sheep flock was fed the high-nitrogen diet, thyroxin and proline levels were significantly reduced in hydrolyzates of epimural bacteria from all parts of the rumen, while low-nitrogen diet significantly increased the concentrations of both given amino acids in comparison with the control. Bartos (1987
Müller, Herbert; Reissner, Michael; Steiner, Walter; Wiesinger, Günter
. The positive atmosphere, the high attendance in the sessions and the lively discussions made the conference a great success and a memorable event. It was pointed out, that Mössbauer spectroscopy is still an interesting and powerful method with great opportunities in the future. Herbert Müller (Secretary) Michael Reissner (Chairman) This book is dedicated to our colleagues Nicol Malcom, who could not come, because he suddenly died a few weeks in advance to the conference and Hercilio Rechenberg, who died on his way home from Vienna. Conference photograph Conference Organisation Local Organizing Committee Reissner Michael (Chairman)Müller Herbert (Conference Secretary) Amthauer Georg Lottermoser WernerSteiner Walter Bauer Ernst Michor Herwig Vogl Gero Bühler-Paschen Silke Müller Martin Waas Monika Grodzicki Michael Redhammer Günther Wiesinger Günter Grössinger Roland Sassik Herbert Hilscher Gerfried Sepiol Bogdan International Programme Committee Amthauer Georg Gütlich Philipp Steiner Walter Baggio-Saitovich Elisa Litterst Fred Jochen Trautwein Alfred Xaver Berry Frank Long Gary Vogl Gero Felner Israel Nagy Denes Lajos Yoshida Yutaka Greneche Jean-Marc Rüffer Rudolf International Advisory Board Alp E ErcanGénin Jean-Marie Baggio-Saitovitch Elisa Greneche Jean-Marc Miglierini Marcel Balogh Judit Grodzicki Michael Musić Svetozar Bender Koch Christian Gütlich Philipp Nagy Dénes Lajos Berry Frank Häggström Lennart Nishida Tetsuaki Brown Dennis Hanzel Darko Pérez Alcázar German Campbell Stewart Hassaan Mohamed Yousri Rüffer Rudolf Carbucicchio Massimo Jumas Jean-Claude Ryan Dominic H Croci Simonetta Kadyrzhanov Kariat Sanchez Francisco Di Naili Katila Toivo Schünemann Volker Elzain Mohamed Kim Chul Sung Stanek Jan Fabris José Domingos Klingelhöfer Göstar Stevens John Felner Israel Langouche Guido Suzdalev Igor P Fern George R Lyubutin Igor S Szymanski Krzysztof Forder Sue D Marco Jose F Waanders Frans Gajbhiye Nandeo Mašlaň Miroslav Yoshida Yutaka
Hales, Matthew; Martens, Wayde; Steinberg, Theodore
. The test systems and experimental results obtained will be presented. 1. Okubo, T., Tsuchida, A., Okuda, T., Fujitsuna, K., Ishikawa, M., Morita, T., Tada, T. , Kinetic Analyses of Colloidal Crystallization in Microgravity -Aircraft Experiments. . Colloids and Surfaces A: Physicochemical and Engineering Aspects, 1999. 153: p. 515-524. 2. Okubo, T., Tsuchida, A., Kobayashi, K., Kuno, A., Morita, T., Fujishima, M., Kohno, Y., Kinetic Study of the Formation Reaction of Colloidal Silica Spheres in Microgravity Using Aircraft. Colloid Polymer Science, 1999. 277(5): p. 474-478. 3. Pienaar, C.L., Chiffoleau, G. J. A., Follens, L. R. A., Martens, J. A., Kirschhock, C. E. A., Steinberg, T. A., Effect of Gravity on the Gelation of Silica Sols. Chem. Mater., 2007. 19(4): p. 660-664. 4. Smith, D.D., et al., Effect of Microgravity on the Growth of Silica Nanostructures. Langmuir, 2000. 16(26): p. 10055-10060. 5. Zhang, X., Johnson, D.P., Manerbino, A.R., Moore, J.J., Schowengerdt, F. , Recent Mi-crogravity Results in the Synthesis of Porous Materials. AIP Conference Proceedings (Space Technology and Applications International Forum-1999, Pt. 1), 1999. 458: p. 88-93. 6. Dunbar, P.B., Bendzko, N.J.,, 1H and 13C NMR observation of the reaction of acetic acid with titanium isopropoxide. Materials Chemistry and Physics, 1999. 59: p. 26-35. 7. Krunks, M., Oja, I., T˜nsuaadu, K., Es-Souni, M., Gruselle, M., Niinistü,. L, Thermoanalytical study of acetylacetonate-modified titanium (iv) isopropoxide as precursor for TiO2 films. Journal of Thermal Analysis and Calorimetry, 2005: p. 483-488. 8. Moran, P.D., Bowmaker, G. A., Cooney, R. P., Vibrational Spectra and Molecular Associa-tion of Titanium Tetraisopropoxide. Inorg. Chem., 1998. 37(1): p. 2741-2748. 9. Somogyvari, A., Serpone, N.,, Evidence for five-coordination in titanium(1V) complexes. A nuclear magnetic resonance investigation. Canadian Journal of Chemistry, 1977. 56: p. 316-319.
Szu, Harold H.
The early vision principle of redundancy reduction of 108 sensor excitations is understandable from computer vision viewpoint toward sparse edge maps. It is only recently derived using a truly unsupervised learning paradigm of artificial neural networks (ANN). In fact, the biological vision, Hubel- Wiesel edge maps, is reproduced seeking the underlying independent components analyses (ICA) among 102 image samples by maximizing the ANN output entropy (partial)H(V)/(partial)[W] equals (partial)[W]/(partial)t. When a pair of newborn eyes or ears meet the bustling and hustling world without supervision, they seek ICA by comparing 2 sensory measurements (x1(t), x2(t))T equalsV X(t). Assuming a linear and instantaneous mixture model of the external world X(t) equals [A] S(t), where both the mixing matrix ([A] equalsV [a1, a2] of ICA vectors and the source percentages (s1(t), s2(t))T equalsV S(t) are unknown, we seek the independent sources
approximately equals [I] where the approximated sign indicates that higher order statistics (HOS) may not be trivial. Without a teacher, the ANN weight matrix [W] equalsV [w1, w2] adjusts the outputs V(t) equals tanh([W]X(t)) approximately equals [W]X(t) until no desired outputs except the (Gaussian) 'garbage' (neither YES '1' nor NO '-1' but at linear may-be range 'origin 0') defined by Gaussian covariance Oja, Bell-Sejnowski, Amari-Cichocki, Cardoso), the LYAPONOV function L(v1,...,vn, w1,...wn,) equals E(v1,...,vn) - H(w1,...wn) is constructed as the HELMHOTZ free energy to prove both convergences of supervised energy E and unsupervised entropy H learning. Consequently, rather