Incorporating complex cells into neural networks for pattern classification

Dans le domaine des neurosciences computationnelles, l'hypothèse a été émise que le système visuel, depuis la rétine et jusqu'au cortex visuel primaire au moins, ajuste continuellement un modèle probabiliste avec des variables latentes, à son flux de perceptions. Ni le modèle exact, ni la méthode exacte utilisée pour l'ajustement ne sont connus, mais les algorithmes existants qui permettent l'ajustement de tels modèles ont besoin de faire une estimation conditionnelle des variables latentes. Cela nous peut nous aider à comprendre pourquoi le système visuel pourrait ajuster un tel modèle; si le modèle est approprié, ces estimé conditionnels peuvent aussi former une excellente représentation, qui permettent d'analyser le contenu sémantique des images perçues. Le travail présenté ici utilise la performance en classification d'images (discrimination entre des types d'objets communs) comme base pour comparer des modèles du système visuel, et des algorithmes pour ajuster ces modèles (vus comme des densités de probabilité) à des images. Cette thèse (a) montre que des modèles basés sur les cellules complexes de l'aire visuelle V1 généralisent mieux à partir d'exemples d'entraînement étiquetés que les réseaux de neurones conventionnels, dont les unités cachées sont plus semblables aux cellules simples de V1; (b) présente une nouvelle interprétation des modèles du système visuels basés sur des cellules complexes, comme distributions de probabilités, ainsi que de nouveaux algorithmes pour les ajuster à des données; et (c) montre que ces modèles forment des représentations qui sont meilleures pour la classification d'images, après avoir été entraînés comme des modèles de probabilités. Deux innovations techniques additionnelles, qui ont rendu ce travail possible, sont également décrites : un algorithme de recherche aléatoire pour sélectionner des hyper-paramètres, et un compilateur pour des expressions mathématiques matricielles, qui peut optimiser ces expressions pour processeur central (CPU) et graphique (GPU).

Is social phobia characterized by a distinct interpersonal pattern?

Traditionnellement, le construit de la phobie sociale a été défini selon une vision intrapersonnelle, en tant que trouble de l’anxiété. Une autre conception se propose de la définir d’un point de vue interpersonnel, comme un pattern global d’autoprotection. L’objectif principal de cette thèse est de tester des hypothèses tirées du modèle interpersonnel de la phobie sociale. Deux études, présentées sous forme d’articles, ont permis d’examiner si des patterns spécifiques d’autoprotection, tels que l’impuissance et la soumission, caractérisent le mode de fonctionnement des phobiques sociaux. Les études ont également évalué si l’autoprotection et l’anxiété sont interreliées. Pour la première étude, les patterns interpersonnels de 132 phobiques sociaux, évalués à l’aide d’une mesure dérivée du Circumplex interpersonnel, ont été comparés à ceux de 85 individus célibataires ayant une dysfonction sexuelle et 105 sujets normaux. La relation entre les patterns d’autoprotection, l’anxiété sociale, la détresse générale et le fonctionnement social a également été examinée chez les phobiques sociaux. La seconde étude a permis d’examiner l’évolution des patterns d’autoprotection ainsi que de l’anxiété sociale, de la détresse générale et du fonctionnement social, chez 85 phobiques sociaux à quatre moments : avant et après un traitement d’approche interpersonnelle, ainsi qu’aux relances de six mois et d’un an. L’étude a également comparé les participants en rémission et ceux satisfaisant les critères de la phobie sociale un an suivant la fin du traitement. Les résultats suggèrent que les patterns d’impuissance et de soumission sont caractéristiques de la phobie sociale. Plus précisément, ces patterns décrivent davantage les comportements des phobiques sociaux plutôt que ceux des groupes de comparaison. De plus, une réduction significative de l’autoprotection a été notée au post-traitement et maintenue jusqu’au suivi d’un an, surtout chez les participants en rémission.En outre, une relation entre l’autoprotection, l’anxiété sociale et la détresse générale a été mise en évidence chez les phobiques sociaux. Une amélioration de l’anxiété, de la détresse subjective et du fonctionnement social cohérente avec la dissolution des patterns d’autoprotection a également été obtenue au post-traitement. En conclusion, les résultats des deux études appuient une conception interpersonnelle de la phobie sociale.

Consumption Expenditure Pattern of Scheduled Caste Households of Kerala: a Study of Idukki District

The present study made an attempt to analyze the socio-economic background and the consumption pattern of scheduled caste households in Idukki district. The objectives of the study are to examine consumption pattern among the scheduled cast population, differences in the average consumption expenditure of different decile groups, consumption expenditure elasticity of items, variations in expenditure of SC households on food, non-food and total expenditure and to examine the association between consumption expenditure and variables such as income, education, occupation and area of residence. The study reveals that the Monthly Per Capita Expenditure of scheduled castes population in rural Kerala is lower than that of the general population. Average household size is higher in rural sector for Scheduled Caste in Kerala as well as all-India. The per capita expenditure of Scheduled Castes of rural Kerala is found to be much lower than that of general population. The study has found that the levels of livings of the Scheduled Castes are far the below the expectations. Large percentage of the Scheduled Caste belongs to the lower income groups. This is due their very low economic status and the consequent employment prospects in low paying occupations. The consumption standards of the majority of Scheduled Castes are found much below that of General population. Effective implementation of the Schemes for their economic upliftment is needed for improving their consumption standards, Minimum Wage Act in the case of agricultural labourers etc. are some of recommendations on the basis of this study.

Array pattern nulling by element position perturbations using a genetic algorithm

A genetic algorithm has been used for null steering in phased and adaptive arrays . It has been shown that it is possible to steer the array null s precisely to the required interference directions and to achieve any prescribed null depths . A comparison with the results obtained from the analytic solution shows the advantages of using the genetic algorithm for null steering in linear array patterns

Comparison Study of Pattern Synthesis Techniques Using Neural Networks

In this paper, a comparison study among three neuralnetwork algorithms for the synthesis of array patterns is presented. The neural networks are used to estimate the array elements' excitations for an arbitrary pattern. The architecture of the neural networks is discussed and simulation results are presented. Two new neural networks, based on radial basis functions (RBFs) and wavelet neural networks (WNNs), are introduced. The proposed networks offer a more efficient synthesis procedure, as compared to other available techniques

STUDIES ON HAEMOLYMPH CONSTITUENTS OF INDOPLANORBIS EXUSTUS (DESHAYES) AND LYMNAEA ACUMINATA (LAMARCK) F.RUFESCENS (GRAY) AND THE EFFECTS OF COPPER ON THE ACTIVITY PATTERN OF SELECTED TRANSAMINASES AND PHOSPHATASES

The problem investigated is on the haematological aspects of two freshwater pulmonate snails, Indoplanorbis exustus (Deshayes),and Lymnaea acuminata f.rufescens (Gray). An important aspect of the present investigation is to emphasize the utilization of freshwater organisms as models for research directed at understanding the basic biomedical problems that remain unresolved. Another aspect is to demonstrate how haemolymph can be treated as a tissue because of late, it has been shown that several parameters of blood can be taken as reliable indicators for diagnostic purposes, and also to monitor environmental pollution. The various haematological parameters studied are total haemocyte number,packed cell volume, haemoglobin, and inorganic as well organic constituents in three size groups of both the snail species. The effect of copper toxicity was measured in terms of total haemocyte count, and the activity pattern of selected phosphatases and transaminases.The study concluded that enzyme activity levels can be taken as reliable indicators to monitor pollution. Age is a factor that determines several of the physiological, biochemical and metabolic activities. This study also indicates that haemolymph can be taken as an organ system to study the various changes taking place at organ systems levels.

STUDIES ON HAEMOLYMPH CONSTITUENTS OF PILA VIRENS (LAMARCK) AND THE EFFECTS OF PESTICIDES ON THE ACTIVITY PATTERN OF SELECTED ENZYMES

The present study is an attempt to understand the physiological responses of a freshwater gastropod, in terms of haematological parameters, in normal conditions as well as in various natural and man made altered conditions of the environment.Pila virens, a freshwater prosobranch,commonly found in paddy fields, ponds, and streams of Kerala is selected as the test animal for the present investigation. Various haemolymph constituents such as total carbohydrate, glycogen, total protein, total lipid, urea,ammonia,sodium,potasium, calcium, and chloride which are directly involved in the control and maintenance of different physiological systems, were analysed in the present study. Selected haematological parameters like total haemocyte number, and packed cell volume were also determined. Besides , the activity pattern of selected haemolymph enzymes such as acid phosphatase (ACP), alkaline phosphatase (ALP),Glutamate-oxaloacetate transaminase (GOT), and glutamate-pyruvate transaminase (GPT), all having diagnostic value in terms of internal defence system and metabolism of the organism, were also studied.

Applications of Fractal Based Structures in Radar Cross Section Reduction

This thesis presents the Radar Cross Section measurements of different geometric structures such as flat plate,cylinder, corner reflector and circular cone loaded with fractal based metallo dielectric structures.Use of different fractal geometris,metallizations of different shapes as well as the frequency tanability is investigated for TE and TM polarization of the incident electromagnetic field.Application of fractal based metallo-dielectric structures results in RCS reduction over a wide range of frequency bands.RCS enhancement of dihedral corner is observed at certain acute and obtuse corner angles.The experimental results are validated using electromagnetic simulation softwares.

Scattering Behaviour of fractal based metallo-dielectric structures

The scattering behaviour of fractal based metallodielectric structures loaded over metallic targets of different shapes such as flat plate, cylinder and dihedral corner reflector are investigated for both TE and TM polarizations of the incident wave. Out of the various fractal structures studied,square Sierpinski carpet structure is found to give backscattering reduction for an appreciable range of frequencies. The frequency of minimum backscattering depends on the geometry of the structure as well as on the thickness of the substrate. This structure when loaded over a dihedral corner reflector is showing an enhancement in RCS for corner angles other than 90◦.

Radar cross-section enhancement of dihedral corner reflector using fractal-based metallo-dielectric structures

Effective use of fractal-based metallo-dielectric structures for enhancing the radar cross-section (RCS) of dihedral corner reflectors is reported. RCS enhancement of about 30 dBsm is obtained for corner reflectors with corner angles other than 90deg. This may find application in remote sensing and synthetic aperture radar.

Strip-loaded hollow dielectric H-plane sectoral horn antennas for square radiation pattern

The radiation characteristics of a new type of hollow dielectric H-plane sectoral horn antenna are presented. Metallic strips of optimum length are loaded on the H-walls of the sectoral horns. The effects of strip loading for producing square patterns in the H plane are discussed.

Fractal based techniques for classification of mammograms and identification of microcalcifications

After skin cancer, breast cancer accounts for the second greatest number of cancer diagnoses in women. Currently the etiologies of breast cancer are unknown, and there is no generally accepted therapy for preventing it. Therefore, the best way to improve the prognosis for breast cancer is early detection and treatment. Computer aided detection systems (CAD) for detecting masses or micro-calcifications in mammograms have already been used and proven to be a potentially powerful tool , so the radiologists are attracted by the effectiveness of clinical application of CAD systems. Fractal geometry is well suited for describing the complex physiological structures that defy the traditional Euclidean geometry, which is based on smooth shapes. The major contribution of this research include the development of • A new fractal feature to accurately classify mammograms into normal and normal (i)With masses (benign or malignant) (ii) with microcalcifications (benign or malignant) • A novel fast fractal modeling method to identify the presence of microcalcifications by fractal modeling of mammograms and then subtracting the modeled image from the original mammogram. The performances of these methods were evaluated using different standard statistical analysis methods. The results obtained indicate that the developed methods are highly beneficial for assisting radiologists in making diagnostic decisions. The mammograms for the study were obtained from the two online databases namely, MIAS (Mammographic Image Analysis Society) and DDSM (Digital Database for Screening Mammography.