New multiple matched filter: design and experimental realization

A method of making a multiple matched filter which allows the recognition of different characters in successive planes in simple conditions is proposed. The generation of the filter is based on recording on the same plate the Fourier transforms of the different patterns to be recognized, each of which is affected by different spherical phase factors because the patterns have been placed at different distances from the lens. This is proved by means of experiments with a triple filter which allows satisfactory recognition of three characters.

Nonlinear filtering in object and Fourier space in a joint transform optical correlator: comparison and experimental realization

The use of different kinds of nonlinear filtering in a joint transform correlator are studied and compared. The study is divided into two parts, one corresponding to object space and the second to the Fourier domain of the joint power spectrum. In the first part, phase and inverse filters are computed; their inverse Fourier transforms are also computed, thereby becoming the reference in the object space. In the Fourier space, the binarization of the power spectrum is realized and compared with a new procedure for removing the spatial envelope. All cases are simulated and experimentally implemented by a compact joint transform correlator.

Chlamydiales and the innate immune response: friend or foe?

Pathogenicity of Chlamydia and Chlamydia-related bacteria could be partially mediated by an enhanced activation of the innate immune response. The study of this host pathogen interaction has proved challenging due to the restricted in vitro growth of these strict intracellular bacteria and the lack of genetic tools to manipulate their genomes. Despite these difficulties, the interactions of Chlamydiales with the innate immune cells and their effectors have been studied thoroughly. This review aims to point out the role of pattern recognition receptors and signal molecules (cytokines, reactive oxygen species) of the innate immune response in the pathogenesis of chlamydial infection. Besides inducing clearance of the bacteria, some of these effectors may be used by the Chlamydia to establish chronic infections or to spread. Thus, the induced innate immune response seems to be variable depending on the species and/or the serovar, making the pattern more complex. It remains crucial to determine the common players of the innate immune response in order to help define new treatment strategies and to develop effective vaccines. The excellent growth in phagocytic cells of some Chlamydia-related organisms such as Waddlia chondrophila supports their use as model organisms to study conserved features important for interactions between the innate immunity and Chlamydia.

Adaptive tracking of EEG oscillations.

Neuronal oscillations are an important aspect of EEG recordings. These oscillations are supposed to be involved in several cognitive mechanisms. For instance, oscillatory activity is considered a key component for the top-down control of perception. However, measuring this activity and its influence requires precise extraction of frequency components. This processing is not straightforward. Particularly, difficulties with extracting oscillations arise due to their time-varying characteristics. Moreover, when phase information is needed, it is of the utmost importance to extract narrow-band signals. This paper presents a novel method using adaptive filters for tracking and extracting these time-varying oscillations. This scheme is designed to maximize the oscillatory behavior at the output of the adaptive filter. It is then capable of tracking an oscillation and describing its temporal evolution even during low amplitude time segments. Moreover, this method can be extended in order to track several oscillations simultaneously and to use multiple signals. These two extensions are particularly relevant in the framework of EEG data processing, where oscillations are active at the same time in different frequency bands and signals are recorded with multiple sensors. The presented tracking scheme is first tested with synthetic signals in order to highlight its capabilities. Then it is applied to data recorded during a visual shape discrimination experiment for assessing its usefulness during EEG processing and in detecting functionally relevant changes. This method is an interesting additional processing step for providing alternative information compared to classical time-frequency analyses and for improving the detection and analysis of cross-frequency couplings.

ZHARP: three-dimensional motion tracking from a single image plane.

Three-dimensional imaging and quantification of myocardial function are essential steps in the evaluation of cardiac disease. We propose a tagged magnetic resonance imaging methodology called zHARP that encodes and automatically tracks myocardial displacement in three dimensions. Unlike other motion encoding techniques, zHARP encodes both in-plane and through-plane motion in a single image plane without affecting the acquisition speed. Postprocessing unravels this encoding in order to directly track the 3-D displacement of every point within the image plane throughout an entire image sequence. Experimental results include a phantom validation experiment, which compares zHARP to phase contrast imaging, and an in vivo study of a normal human volunteer. Results demonstrate that the simultaneous extraction of in-plane and through-plane displacements from tagged images is feasible.

Automatic selection of tidal volume, respiratory frequency and minute ventilation in intubated ICU patients as start up procedure for closed-loop controlled ventilation.

OBJECTIVE: Before a patient can be connected to a mechanical ventilator, the controls of the apparatus need to be set up appropriately. Today, this is done by the intensive care professional. With the advent of closed loop controlled mechanical ventilation, methods will be needed to select appropriate start up settings automatically. The objective of our study was to test such a computerized method which could eventually be used as a start-up procedure (first 5-10 minutes of ventilation) for closed-loop controlled ventilation. DESIGN: Prospective Study. SETTINGS: ICU's in two adult and one children's hospital. PATIENTS: 25 critically ill adult patients (age > or = 15 y) and 17 critically ill children selected at random were studied. INTERVENTIONS: To stimulate 'initial connection', the patients were disconnected from their ventilator and transiently connected to a modified Hamilton AMADEUS ventilator for maximally one minute. During that time they were ventilated with a fixed and standardized breath pattern (Test Breaths) based on pressure controlled synchronized intermittent mandatory ventilation (PCSIMV). MEASUREMENTS AND MAIN RESULTS: Measurements of airway flow, airway pressure and instantaneous CO2 concentration using a mainstream CO2 analyzer were made at the mouth during application of the Test-Breaths. Test-Breaths were analyzed in terms of tidal volume, expiratory time constant and series dead space. Using this data an initial ventilation pattern consisting of respiratory frequency and tidal volume was calculated. This ventilation pattern was compared to the one measured prior to the onset of the study using a two-tailed paired t-test. Additionally, it was compared to a conventional method for setting up ventilators. The computer-proposed ventilation pattern did not differ significantly from the actual pattern (p > 0.05), while the conventional method did. However the scatter was large and in 6 cases deviations in the minute ventilation of more than 50% were observed. CONCLUSIONS: The analysis of standardized Test Breaths allows automatic determination of an initial ventilation pattern for intubated ICU patients. While this pattern does not seem to be superior to the one chosen by the conventional method, it is derived fully automatically and without need for manual patient data entry such as weight or height. This makes the method potentially useful as a start up procedure for closed-loop controlled ventilation.

Persistent infection of arenaviruses : molecular mechanisms and pathogenesis

Arenaviruses are enveloped negative single strand RNA viruses that include a number of important human pathogens. The most prevalent human pathogen among the arenaviruses is the Old World arenavirus Lassa virus (LASV) which is endemic in West Africa from Senegal to Cameroon. LASV is the etiologic agent of a severe viral hemorrhagic fever named Lassa fever whose mortality rate can reach 30% in hospitalized patients. One of the hallmarks of fatal arenavirus infection in humans is the absence of an effective innate and adaptive immune response. In nature, arenaviruses are carried by rodents which represent the natural reservoirs as well as the vectors for transmission. In their natural rodent reservoir, arenaviruses have the ability to establish persistent infection without any overt signs and symptoms of pathology. We believe that the modulation of the host cell's innate immunity by arenaviruses is a key determinant for persistence in the natural host and for the pathogenesis in man. In this thesis, we studied the interaction of arenaviruses with two main branches of the host's innate anti-viral defense, the type I interferon (IFN) system and virus-induced mitochondrial apoptosis. The arenavirus nucleoprotein (NP) is responsible for the anti-IFN activity of arenaviruses. Specifically, NP blocks the activation and the nuclear translocation of the transcription factor interferon regulatory factor 3 (IRF3) which leads to type I IFN production. LASV and the prototypic arenavirus lymphocytic choriomeningitis virus (LCMV) NPs contain a 3'-5'exoribonuclease domain in the C terminal part that has been linked to the anti-IFN activity of NP. In the first project, we sought to identify cellular component(s) of the type I IFN induction pathway targeted by the viral NP. Our study revealed that LCMV NP prevents the activation of IRF3 by blocking phosphorylation of the transcription factor. We found that LCMV NP specifically targets the IRF-activating kinase IKKs, and this specific binding is conserved within the Arenaviridae. We could also demonstrate that LCMV NP associates with the kinase domain of IKKs involving NP's C-terminal region. Lastly, we showed that the binding of LCMV NP inhibits the kinase activity of IKKs. This study allowed the discovery of a new cellular interacting partner of arenavirus NP. This newly described association may play a role in the anti-IFN activity of arenaviruses but potentially also in other aspects of arenavirus infection. For the second project, we investigated the ability of arenaviruses to avoid and/or suppress mitochondrial apoptosis. As persistent viruses, arenaviruses evolved a "hit and stay" survival strategy where the apoptosis of the host cell would be deleterious. We found that LCMV does not induce mitochondrial apoptosis at any time during infection. Specifically, no caspase activity, no cytochrome c release from the mitochondria as well as no cleavage of poly (ADP-ribose) polymerase (PARP) were detected during LCMV infection. Interestingly, we found that virus-induced mitochondrial apoptosis remains fully functional in LCMV infected cells, while the induction of type IIFN is blocked. Since both type IIFN production and virus- induced mitochondrial apoptosis critically depend on the pattern recognition receptor (PRR) RIG-I, we examined the role of RIG-I in apoptosis in LCMV infected cells. Notably, virus- induced mitochondrial apoptosis in LCMV infected cells was found to be independent of RIG- I and MDA5, but still depended on MAVS. Our study uncovered a novel mechanism by which arenaviruses alter the host cell's pro-apoptotic signaling pathway. This might represent a strategy arenaviruses developed to maintain this branch of the innate anti-viral defense in absence of type I IFN response. Taken together, these results allow a better understanding of the interaction of arenaviruses with the host cell's innate immunity, contributing to our knowledge about pathogenic properties of these important viruses. A better comprehension of arenavirus virulence may open new avenues for vaccine development and may suggest new antiviral targets for therapeutic intervention against arenavirus infections. - Les arenavirus sont des virus enveloppés à ARN simple brin qui comportent un grand nombre de pathogènes humains. Le pathogène humain le plus important parmi les arenavirus est le virus de Lassa qui est endémique en Afrique de l'Ouest, du Sénégal au Cameroun. Le virus de Lassa est l'agent étiologique d'une fièvre hémorragique sévère appelée fièvre de Lassa, et dont le taux de mortalité peut atteindre 30% chez les patients hospitalisés. L'une des caractéristiques principales des infections fatales à arenavirus chez l'Homme est l'absence de réponse immunitaire innée et adaptative. Dans la nature, les arenavirus sont hébergés par différentes espèces de rongeur, qui représentent à la fois les réservoirs naturels et les vecteurs de transmission des arenavirus. Dans leur hôte naturel, les arenavirus ont la capacité d'établir une infection persistante sans symptôme manifeste d'une quelconque pathologie. Nous pensons que la modulation de système immunitaire inné de la cellule hôte par les arenavirus est un paramètre clé pour la persistance au sein de l'hôte naturel, ainsi que pour la pathogenèse chez l'Homme. L'objectif de cette thèse était d'étudier l'interaction des arenavirus avec deux branches essentielles de la défense antivirale innée de la cellule hôte, le système interféron (IFN) de type I et l'apoptose. La nucléoprotéine virale (NP) est responsable de l'activité anti-IFN des arenavirus. Plus spécifiquement, la NP bloque 1'activation et la translocation nucléaire du facteur de transcription IRF3 qui conduit à la production des IFNs de type I. La NP du virus de Lassa et celle du virus de la chorioméningite lymphocytaire (LCMV), l'arénavirus prototypique, possèdent dans leur extrémité C-terminale un domaine 3'-5' exoribonucléase qui a été associé à l'activité anti-IFN de ces protéines. Dans un premier projet, nous avons cherché à identifier des composants cellulaires de la cascade de signalisation induisant la production d'IFNs de type I qui pourraient être ciblés par la NP virale. Nos recherches ont révélé que la NP de LCMV empêche 1'activation d'IRF3 en bloquant la phosphorylation du facteur de transcription. Nous avons découvert que la NP de LCMV cible spécifiquement la kinase IKKe, et que cette interaction spécifique est conservée à travers la famille des Arenaviridae. Notre étude a aussi permis de démontrer que la NP de LCMV interagit avec le domaine kinase d'IKKe et que l'extrémité C-terminale de la NP est impliquée. Pour finir, nous avons pu établir que l'association avec la NP de LCMV inhibe l'activité kinase d'IKKe. Cette première étude présente la découverte d'un nouveau facteur cellulaire d'interaction avec la NP des arenavirus. Cette association pourrait jouer un rôle dans l'activité anti-IFN des arénavirus, mais aussi potentiellement dans d'autres aspects des infections à arénavirus. Pour le second projet, nous nous sommes intéressés à la capacité des arénavirus à éviter et/ou supprimer l'apoptose mitochondriale. En tant que virus persistants, les arénavirus ont évolué vers une stratégie de survie "hit and stay" pour laquelle l'apoptose de la cellule hôte serait néfaste. Nous avons observé qu'à aucun moment durant l'infection LCMV n'induit l'apoptose mitochondriale. Spécifiquement, aucune activité de caspase, aucune libération mitochondriale de cytochrome c ainsi qu'aucun clivage de la polymerase poly(ADP-ribose) (PARP) n'a été détecté pendant l'infection à LCMV. Il est intéressant de noter que l'apoptose mitochondriale induite par les virus reste parfaitement fonctionnelle dans les cellules infectées par LCMV, alors que l'induction de la réponse IFN de type I est bloquée dans les mêmes cellules. La production des IFNs de type I et l'apoptose mitochondriale induite par les virus dépendent toutes deux du récepteur de reconnaissance de motifs moléculaires RIG-I. Nous avons, par conséquent, investigué le rôle de RIG-I dans l'apoptose qui a lieu dans les cellules infectées par LCMV lorsqu'on les surinfecte avec un autre virus pro-apoptotique. En particulier, l'apoptose mitochondriale induite par les surinfections s'est révélée indépendante de RIG-I et MDA5, mais dépendante de MAVS dans les cellules précédemment infectées par LCMV. Notre étude démontre ainsi l'existence d'un nouveau mécanisme par lequel les arénavirus altèrent la cascade de signalisation pro-apoptotique de la cellule hôte. Il est possible que les arénavirus aient développé une stratégie permettant de maintenir fonctionnelle cette branche de la défense antivirale innée en l'absence de réponse IFN de type I. En conclusion, ces résultats nous amènent à mieux comprendre l'interaction des arénavirus avec l'immunité innée de la cellule hôte, ce qui contribue aussi à améliorer notre connaissance des propriétés pathogéniques de ces virus. Une meilleure compréhension des facteurs de virulence des arénavirus permet, d'une part, le développement de vaccins et peut, d'autre part, servir de base pour la découverte de nouvelles cibles thérapeutiques utilisées dans le traitement des infections à arénavirus.

Efficient total variation algorithm for fetal brain MRI reconstruction.

Fetal MRI reconstruction aims at finding a high-resolution image given a small set of low-resolution images. It is usually modeled as an inverse problem where the regularization term plays a central role in the reconstruction quality. Literature has considered several regularization terms s.a. Dirichlet/Laplacian energy, Total Variation (TV)- based energies and more recently non-local means. Although TV energies are quite attractive because of their ability in edge preservation, standard explicit steepest gradient techniques have been applied to optimize fetal-based TV energies. The main contribution of this work lies in the introduction of a well-posed TV algorithm from the point of view of convex optimization. Specifically, our proposed TV optimization algorithm or fetal reconstruction is optimal w.r.t. the asymptotic and iterative convergence speeds O(1/n2) and O(1/√ε), while existing techniques are in O(1/n2) and O(1/√ε). We apply our algorithm to (1) clinical newborn data, considered as ground truth, and (2) clinical fetal acquisitions. Our algorithm compares favorably with the literature in terms of speed and accuracy.

Clinical and electrophysiological recovery in Leber hereditary optic neuropathy with G3460A mutation.

To report a case of clinical and electrophysiological recovery in Leber hereditary optic neuropathy (LHON) with G3460A Mutation. A 10-year-old boy with a three-month history of painless bilateral sequential visual loss upon presentation underwent visual acuity (diminished), anterior and posterior segment examination (normal), fluorescein angiography (normal), Goldman kinetic perimetry (bilateral central scotomata), genetic (a point G3460A mutation) and electrophysiological investigation (undetectable pattern visual evoked potentials (VEP); low amplitude, broadened and reduced flash VEPs and loss of the N95 component in the pattern electroretinograms). Diagnosis of LHON was made. Eighteen months later vision and electrophysiological tests results began spontaneously improving. Kinetic perimetry revealed reduced density and size of scotomata. Two years later, there had been further electrophysiological improvement. This report describes both clinical and electrophysiological improvement in LHON with G3460A mutation.

Segmentation of foot and ankle complex based on kinematic criteria.

Although various foot models were proposed for kinematics assessment using skin makers, no objective justification exists for the foot segmentations. This study proposed objective kinematic criteria to define which foot joints are relevant (dominant) in skin markers assessments. Among the studied joints, shank-hindfoot, hindfoot-midfoot and medial-lateral forefoot joints were found to have larger mobility than flexibility of their neighbour bonesets. The amplitude and pattern consistency of these joint angles confirmed their dominancy. Nevertheless, the consistency of the medial-lateral forefoot joint amplitude was lower. These three joints also showed acceptable sensibility to experimental errors which supported their dominancy. This study concluded that to be reliable for assessments using skin markers, the foot and ankle complex could be divided into shank, hindfoot, medial forefoot, lateral forefoot and toes. Kinematics of foot models with more segments must be more cautiously used.

Automated identification of minimal myocardial motion for improved image quality on MR angiography at 3 T

OBJECTIVE: Imaging during a period of minimal myocardial motion is of paramount importance for coronary MR angiography (MRA). The objective of our study was to evaluate the utility of FREEZE, a custom-built automated tool for the identification of the period of minimal myocardial motion, in both a moving phantom at 1.5 T and 10 healthy adults (nine men, one woman; mean age, 24.9 years; age range, 21-32 years) at 3 T. CONCLUSION: Quantitative analysis of the moving phantom showed that dimension measurements approached those obtained in the static phantom when using FREEZE. In vitro, vessel sharpness, signal-to-noise ratio (SNR), and contrast-to-noise ratio (CNR) were significantly improved when coronary MRA was performed during the software-prescribed period of minimal myocardial motion (p < 0.05). Consistent with these objective findings, image quality assessments by consensus review also improved significantly when using the automated prescription of the period of minimal myocardial motion. The use of FREEZE improves image quality of coronary MRA. Simultaneously, operator dependence can be minimized while the ease of use is improved.

The immunological, environmental, and phylogenetic perpetrators of metastatic leishmaniasis.

Cutaneous leishmaniases have persisted for centuries as chronically disfiguring parasitic infections affecting millions of people across the subtropics. Symptoms range from the more prevalent single, self-healing cutaneous lesion to a persistent, metastatic disease, where ulcerations and granulomatous nodules can affect multiple secondary sites of the skin and delicate facial mucosa, even sometimes diffusing throughout the cutaneous system as a papular rash. The basis for such diverse pathologies is multifactorial, ranging from parasite phylogeny to host immunocompetence and various environmental factors. Although complex, these pathologies often prey on weaknesses in the innate immune system and its pattern recognition receptors. This review explores the observed and potential associations among the multifactorial perpetrators of infectious metastasis and components of the innate immune system.

Machine learning for geospatial data : algorithms, software tools and case studies

Résumé Cette thèse est consacrée à l'analyse, la modélisation et la visualisation de données environnementales à référence spatiale à l'aide d'algorithmes d'apprentissage automatique (Machine Learning). L'apprentissage automatique peut être considéré au sens large comme une sous-catégorie de l'intelligence artificielle qui concerne particulièrement le développement de techniques et d'algorithmes permettant à une machine d'apprendre à partir de données. Dans cette thèse, les algorithmes d'apprentissage automatique sont adaptés pour être appliqués à des données environnementales et à la prédiction spatiale. Pourquoi l'apprentissage automatique ? Parce que la majorité des algorithmes d'apprentissage automatiques sont universels, adaptatifs, non-linéaires, robustes et efficaces pour la modélisation. Ils peuvent résoudre des problèmes de classification, de régression et de modélisation de densité de probabilités dans des espaces à haute dimension, composés de variables informatives spatialisées (« géo-features ») en plus des coordonnées géographiques. De plus, ils sont idéaux pour être implémentés en tant qu'outils d'aide à la décision pour des questions environnementales allant de la reconnaissance de pattern à la modélisation et la prédiction en passant par la cartographie automatique. Leur efficacité est comparable au modèles géostatistiques dans l'espace des coordonnées géographiques, mais ils sont indispensables pour des données à hautes dimensions incluant des géo-features. Les algorithmes d'apprentissage automatique les plus importants et les plus populaires sont présentés théoriquement et implémentés sous forme de logiciels pour les sciences environnementales. Les principaux algorithmes décrits sont le Perceptron multicouches (MultiLayer Perceptron, MLP) - l'algorithme le plus connu dans l'intelligence artificielle, le réseau de neurones de régression généralisée (General Regression Neural Networks, GRNN), le réseau de neurones probabiliste (Probabilistic Neural Networks, PNN), les cartes auto-organisées (SelfOrganized Maps, SOM), les modèles à mixture Gaussiennes (Gaussian Mixture Models, GMM), les réseaux à fonctions de base radiales (Radial Basis Functions Networks, RBF) et les réseaux à mixture de densité (Mixture Density Networks, MDN). Cette gamme d'algorithmes permet de couvrir des tâches variées telle que la classification, la régression ou l'estimation de densité de probabilité. L'analyse exploratoire des données (Exploratory Data Analysis, EDA) est le premier pas de toute analyse de données. Dans cette thèse les concepts d'analyse exploratoire de données spatiales (Exploratory Spatial Data Analysis, ESDA) sont traités selon l'approche traditionnelle de la géostatistique avec la variographie expérimentale et selon les principes de l'apprentissage automatique. La variographie expérimentale, qui étudie les relations entre pairs de points, est un outil de base pour l'analyse géostatistique de corrélations spatiales anisotropiques qui permet de détecter la présence de patterns spatiaux descriptible par une statistique. L'approche de l'apprentissage automatique pour l'ESDA est présentée à travers l'application de la méthode des k plus proches voisins qui est très simple et possède d'excellentes qualités d'interprétation et de visualisation. Une part importante de la thèse traite de sujets d'actualité comme la cartographie automatique de données spatiales. Le réseau de neurones de régression généralisée est proposé pour résoudre cette tâche efficacement. Les performances du GRNN sont démontrées par des données de Comparaison d'Interpolation Spatiale (SIC) de 2004 pour lesquelles le GRNN bat significativement toutes les autres méthodes, particulièrement lors de situations d'urgence. La thèse est composée de quatre chapitres : théorie, applications, outils logiciels et des exemples guidés. Une partie importante du travail consiste en une collection de logiciels : Machine Learning Office. Cette collection de logiciels a été développée durant les 15 dernières années et a été utilisée pour l'enseignement de nombreux cours, dont des workshops internationaux en Chine, France, Italie, Irlande et Suisse ainsi que dans des projets de recherche fondamentaux et appliqués. Les cas d'études considérés couvrent un vaste spectre de problèmes géoenvironnementaux réels à basse et haute dimensionnalité, tels que la pollution de l'air, du sol et de l'eau par des produits radioactifs et des métaux lourds, la classification de types de sols et d'unités hydrogéologiques, la cartographie des incertitudes pour l'aide à la décision et l'estimation de risques naturels (glissements de terrain, avalanches). Des outils complémentaires pour l'analyse exploratoire des données et la visualisation ont également été développés en prenant soin de créer une interface conviviale et facile à l'utilisation. Machine Learning for geospatial data: algorithms, software tools and case studies Abstract The thesis is devoted to the analysis, modeling and visualisation of spatial environmental data using machine learning algorithms. In a broad sense machine learning can be considered as a subfield of artificial intelligence. It mainly concerns with the development of techniques and algorithms that allow computers to learn from data. In this thesis machine learning algorithms are adapted to learn from spatial environmental data and to make spatial predictions. Why machine learning? In few words most of machine learning algorithms are universal, adaptive, nonlinear, robust and efficient modeling tools. They can find solutions for the classification, regression, and probability density modeling problems in high-dimensional geo-feature spaces, composed of geographical space and additional relevant spatially referenced features. They are well-suited to be implemented as predictive engines in decision support systems, for the purposes of environmental data mining including pattern recognition, modeling and predictions as well as automatic data mapping. They have competitive efficiency to the geostatistical models in low dimensional geographical spaces but are indispensable in high-dimensional geo-feature spaces. The most important and popular machine learning algorithms and models interesting for geo- and environmental sciences are presented in details: from theoretical description of the concepts to the software implementation. The main algorithms and models considered are the following: multi-layer perceptron (a workhorse of machine learning), general regression neural networks, probabilistic neural networks, self-organising (Kohonen) maps, Gaussian mixture models, radial basis functions networks, mixture density networks. This set of models covers machine learning tasks such as classification, regression, and density estimation. Exploratory data analysis (EDA) is initial and very important part of data analysis. In this thesis the concepts of exploratory spatial data analysis (ESDA) is considered using both traditional geostatistical approach such as_experimental variography and machine learning. Experimental variography is a basic tool for geostatistical analysis of anisotropic spatial correlations which helps to understand the presence of spatial patterns, at least described by two-point statistics. A machine learning approach for ESDA is presented by applying the k-nearest neighbors (k-NN) method which is simple and has very good interpretation and visualization properties. Important part of the thesis deals with a hot topic of nowadays, namely, an automatic mapping of geospatial data. General regression neural networks (GRNN) is proposed as efficient model to solve this task. Performance of the GRNN model is demonstrated on Spatial Interpolation Comparison (SIC) 2004 data where GRNN model significantly outperformed all other approaches, especially in case of emergency conditions. The thesis consists of four chapters and has the following structure: theory, applications, software tools, and how-to-do-it examples. An important part of the work is a collection of software tools - Machine Learning Office. Machine Learning Office tools were developed during last 15 years and was used both for many teaching courses, including international workshops in China, France, Italy, Ireland, Switzerland and for realizing fundamental and applied research projects. Case studies considered cover wide spectrum of the real-life low and high-dimensional geo- and environmental problems, such as air, soil and water pollution by radionuclides and heavy metals, soil types and hydro-geological units classification, decision-oriented mapping with uncertainties, natural hazards (landslides, avalanches) assessments and susceptibility mapping. Complementary tools useful for the exploratory data analysis and visualisation were developed as well. The software is user friendly and easy to use.

Classificació i reconeixement de vídeo

El principal objectiu d’aquest projecte és aconseguir classificar diferents vídeos d’esports segons la seva categoria. Els cercadors de text creen un vocabulari segons el significat de les diferents paraules per tal de poder identificar un document. En aquest projecte es va fer el mateix però mitjançant paraules visuals. Per exemple, es van intentar englobar com a una única paraula les diferents rodes que apareixien en els cotxes de rally. A partir de la freqüència amb què apareixien les paraules dels diferents grups dins d’una imatge vàrem crear histogrames de vocabulari que ens permetien tenir una descripció de la imatge. Per classificar un vídeo es van utilitzar els histogrames que descrivien els seus fotogrames. Com que cada histograma es podia considerar un vector de valors enters vàrem optar per utilitzar una màquina classificadora de vectors: una Support vector machine o SVM