Das Ziel dieser Arbeit ist die Formulierung eines Ansatzes zum maschinellen Lernen von Sprachmorphologie, in dem letztere als Zeichenkettentransformationen auf ganzen Wörtern, und nicht als Zerlegung von Wörtern in kleinere stukturelle Einheiten, modelliert wird. Der Beitrag besteht aus zwei wesentlichen Teilen: zum einen wird ein Rechenmodell formuliert, in dem morphologische Regeln als Funktionen auf Zeichenketten definiert sind. Solche Funktionen lassen sich leicht zu endlichen Transduktoren übersetzen, was eine solide algorithmische Grundlage für den Ansatz liefert. Zum anderen wird ein statistisches Modell für Graphen von Wortab\-leitungen eingeführt. Die Inferenz in diesem Modell erfolgt mithilfe des Monte Carlo Expectation Maximization-Algorithmus und die Erwartungswerte über Graphen werden durch einen Metropolis-Hastings-Sampler approximiert. Das Modell wird auf einer Reihe von praktischen Aufgaben evaluiert: Clustering flektierter Formen, Lernen von Lemmatisierung, Vorhersage von Wortart für unbekannte Wörter, sowie Generierung neuer Wörter.

<p>The fast advancing of high-throughput technology has reinforced the biomedical research ecosystem with highly scaled and commercialized data acquisition standards, which provide us with unprecedented opportunity to interrogate biology in novel and creative ways. However, unraveling the high dimensional data in practice is difficult due to the following challenges: 1) how to handle outlier and data contaminations; 2) how to address the curse of dimensionality; 3) how to utilize occasionally provided auxiliary information such as an external phenotype observation or spatial coordinate; 4) how to derive the unknown non-linear relationship between observed data and underlying mechanisms in complex biological system such as human metabolic network. </p> <p><br></p> <p>In sight of the above challenges, this thesis majorly focused on two research directions, for which we have proposed a series of statistical learning and AI-empowered systems biology models. This thesis separates into two parts. The first part focuses on identifying latent low dimensional subspace in high dimensional biomedical data. Firstly, we proposed CAT method which is a robust mixture regression method to detect outliers and estimate parameter simultaneously. Then, we proposed CSMR method in studying the heterogeneous relationship between high dimensional genetic features and a phenotype with penalized mixture regression. At last, we proposed SRMR which investigate mixture linear relationship over spatial domain. The second part focuses on studying the non-linear relationship for human metabolic flux estimation in complex biological system.  We proposed the first method in this domain that can robustly estimate flux distribution of a metabolic network at the resolution of individual cells.</p>

Cette thèse s'inscrit dans le cadre de la conception de véhicules autonomes, et plus spécifiquement de la vérification de contrôleurs de tels véhicules. Nos contributions à la résolution de ce problème sont les suivantes : (1) fournir une syntaxe et une sémantique pour un modèle de systèmes hybrides, (2) étendre les fonctionnalités du model checker statistique Cosmos à ce modèle et (3) valider empiriquement la pertinence de notre approche sur des cas d'étude typiques du véhicule autonome.Nous avons choisi de combiner le modèle des réseaux de Petri stochastiques de haut niveau (qui était le formalisme d'entrée de Cosmos) avec le formalisme d'entrée de Simulink afin d'atteindre un pouvoir d'expression suffisant. En effet Simulink est très largement utilisé dans le domaine automobile et de nombreux contrôleurs sont spécifiés avec cet outil. Or Simulink n'a pas de sémantique formellement définie. Ceci nous a conduit à concevoir une telle sémantique en deux temps : tout d'abord en introduisant une sémantique dite exacte mais qui n'est pas opérationnelle puis en la complétant par une sémantique approchée intégrant le facteur d'approximation recherché.Afin de combiner le modèle à événements discrets des réseaux de Petri et le modèle continu spécifié en Simulink, nous avons proposé au niveau syntaxique une interfacereposant sur de nouveaux types de transitions et au niveau sémantique une extension de la boucle de simulation. L'évaluation de ce nouveau formalisme a été entièrement implémentée dans Cosmos.Grace à ce nouveau formalisme, nous avons développé et étudié les deux cas d'étude suivants : d'une part une circulation dense sur une section d'autoroute et d'autre part l'insertion du véhicule dans une voie rapide. L'analyse des modélisations correspondantes a démontré la pertinence de notre approche. / This thesis takes place in the context of autonomous vehicle design, and concerns more specifically the verification of controllers of such vehicles. Our contributions are the following: (1) give a syntax and a semantics for a hybrid system model, (2) extend the capacities of the model-checker Cosmos to that kind of models, and (3) empirically confirm the relevance of our approach on typical case studies handling autonomous vehicles.We chose to combine high-level stochastic Petri nets (which is the input formalism of Cosmos) with the input formalism of Simulink, to obtain an adequate expressive power. Indeed, Simulink is largely used in the automotive industry and numerous controllers have been specified using this tool. However, there is no formal semantics for Simulink, which lead us to define such a semantics in two steps:first, we propose an exact (but not operational) semantics, then we complete it by an approximate semantics that includes the targeted approximation level.In order to combine the discrete event model of Petri nets and the continous model specified in Simulink, we define a syntactic interface that relies on new transition types; its semantics consists of an extension of the simulation loop. The evaluation of this new formalism has been entirely implemented into Cosmos.Using this new formalism, we have designed and studied the two following case studies: on one hand, a heavy traffic on a motorway segment, and on the other hand the insertion of a vehicle into a motorway. Our approach has been validated by the analysis of the corresponding models.

京都大学 / 0048 / 新制・課程博士 / 博士(工学) / 甲第18311号 / 工博第3903号 / 新制||工||1599(附属図書館) / 31169 / 京都大学大学院工学研究科化学工学専攻 / (主査)教授 長谷部 伸治, 教授 大嶋 正裕, 教授 宮原 稔 / 学位規則第4条第1項該当 / Doctor of Philosophy (Engineering) / Kyoto University / DFAM

京都大学 / 0048 / 新制・課程博士 / 博士(生命科学) / 甲第20531号 / 生博第373号 / 新制||生||49(附属図書館) / 京都大学大学院生命科学研究科高次生命科学専攻 / (主査)教授 松田 道行, 教授 上村 匡, 教授 見学 美根子 / 学位規則第4条第1項該当 / Doctor of Philosophy in Life Sciences / Kyoto University / DFAM

No description available.

The purposes of this study were to develop a notational system to evaluate passing, dribbling, first touch, and individual defensive skills as they relate to success during women's soccer games and to develop a statistical model to weight the importance of each skill on creating scoring opportunities. Sequences of skills in 10 Division I intercollegiate women's soccer games were coded using well defined performance scores and outcomes. The notational analysis system was highly reliable as demonstrated by high test-retest Spearman's correlations (>0.98) between the first and second notation of 3 games for all four skills. The importance scores calculated from a Bayesian model demonstrated that dribbling (0.0127) was the most important skill on creating scoring opportunities, followed by first touch (0.0079), passing (0.0075), and individual defense (0.0050). The notational system developed by this study provides coaches with reliable and objective information in order to improve the specificity of practices and to prepare individuals for optimal performance.

This report shows that a reliable motion detection is needed to make an accurate prediction of future activity. Several experiments are carried out to obtain information about the object ́s behaviour and the best settings for the motion detection. A moving object is captured using two cameras, for two image sequences, and motion detection is applied to the stereoscopic data. Background subtraction algorithm followed by image segmentation algorithm, morphology algorithm, and blob analy- sis are performed on the images to find the coordinates for the centroid of the moving object. Two models are created to make a statistical inter- pretation of the data: one model for the height over the width and one statistical model for the distance between the cameras and the moving object over the width. The mean and standard deviation values are calculated to make a reliable interpretation of the captured images and the moving object. The Kalman filter is used for the prediction of future activity. The filters of the statistical models are trained with the first coordinates of the detected balls, and the next coordinates are predicted.

Face à l’augmentation de la mobilité, les politiques de développement durable cherchent à optimiser l’utilisation des infrastructures de transport existantes. En particulier, les systèmes d’information du trafic à large échelle ont le potentiel d’optimiser l’utilisation du réseau de transport. Ils doivent fournir aux usagers une information fiable en temps réel leur permettant d’optimiser leurs choix d’itinéraires. Ils peuvent également servir d’outils d’aide à la décision pour les gestionnaires du réseau. La thèse étudie comment l’émergence des services Internet sur les téléphones portables et la rapide prolifération des systèmes de géolocalisation permet le développement de nouveaux services d’estimation et d’information des conditions de trafic en réseau urbain. L’utilisation des données provenant de véhicules traceurs nécessite le développement de modèles et d’algorithmes spécifiques, afin d’extraire l’information de ces données qui ne sont envoyées, jusqu’à présent, que par une faible proportion des véhicules circulant sur le réseau et avec une fréquence faible. La variabilité des conditions de circulations, due à la présence de feux de signalisation, motive une approche statistique de la dynamique du trafic, tout en intégrant les principes physiques hydrodynamiques (formation et dissolution de files d’attentes horizontales). Ce modèle statistique permet d’intégrer de façon robuste les données bruitées envoyées par les véhicules traceurs, d’estimer les paramètres physiques caractérisant la dynamique du trafic et d’obtenir l’expression paramétrique de la loi de probabilité des temps de parcours entre deux points quelconques du réseau. La thèse s’appuie sur les données et les infrastructures développées par le projet Mobile Millennium à l’Université de Californie, Berkeley pour valider les modèles et algorithmes proposés. Les résultats soulignent l’importance du développement de modèles statistiques et d’algorithmes adaptés aux données disponibles pour développer un système opérationnel d’estimation du trafic à large échelle / Sustainable mobility development requires the optimization of existing transportation infrastructure. In particular, ubiquitous traffic information systems have the potential to optimize the use of the transportation network. The system must provide accurate and reliable traffic information in real-time to optimize mobility choices. Successful implementations are also valuable tools for traffic management agencies. The thesis studies how the emergence of Internet services and location based services on mobile devices enable the development of novel Intelligent Transportation Systems which estimate and broadcast traffic conditions in arterial networks. Sparsely sampled probe data is the main source of arterial traffic data with the prospect of broad coverage in the near future. The small number of vehicles that report their position at a given time and the low sampling frequency require specific models and algorithms to extract valuable information from the available data. On the one hand, the variability of traffic conditions in urban networks, caused mainly by the presence of traffic lights, motivates a statistical approach of arterial traffic dynamics. On the other hand, an accurate modeling of the physics of arterial traffic from hydrodynamic theory (formation and dissolution of horizontal queues) ensures the physical validity of the model. The thesis proposes to integrate the dynamical model of arterial traffic in a statistical framework to integrate noisy measurements from probe vehicle data and estimate physical parameters, which characterize the traffic dynamics. In particular, the thesis derives and estimates the probability distributions of vehicle location and of travel time between arbitrary locations. The thesis leverages the data and the infrastructure developed by the Mobile Millennium project at the University of California, Berkeley to validate the models and algorithms. The results underline the importance to design statistical models for sparsely sampled probe vehicle data in order to develop the next generation of operation large-scale traffic information systems

Lipídios em solução aquosa formam uma variedade de estruturas diferentes que incluem monocamadas de surfactantes na interface água-ar, conhecidas como monocamadas de Langmuir, como também vesículas unilamelares ou plurilamelares no interior da solução. Sob variação de temperatura, estas estruturas apresentam diferentes fases, observadas através de calorimetria ou variação isotérmica de pressão lateral. Entre as fases apresentadas por estas estruturas, as duas mais importantes se diferenciam pela ordem das cadeias lipídicas. Entendemos que do ponto de vista das fases termodinâmicas, simplificado e qualitativo, monocamadas de Langmuir e bicamadas lipídicas constituem o mesmo sistema físico sob vínculos diferentes. Neste trabalho, desenvolvemos um modelo estatístico para o estudo da transição ordem-desordem destes sistemas, que inclui flutuações de densidade, estas ausentes no modelo de Doniach, de 1980, a base para muitos estudos teóricos para transições de fase de sistemas lipídicos. Flutuações de densidade são fundamentais na descrição de vesículas lipídicas carregadas, compostas de surfactante cuja cabeça polar se dissocia em água. O estudo em laboratório das propriedades térmicas e estruturais de membranas artificiais de lipídios carregados _e relativamente recente, e foi desenvolvido em grande parte no Laboratório de Biofísica do IFUSP. Tais membranas apresentam comportamento distinto das membranas neutras, notoriamente influenciado pela concentração de sal na solução. Isto motivou o desenvolvimento de uma segunda versão do modelo, na qual passamos a descrever a cabeça polar do lipídio em termos de um par de cargas opostas, sendo que a camada lipídica foi acoplada ao modelo primitivo restrito na rede, que desempenha o papel da solução salina. O primeiro modelo foi estudado por aproximação de campo médio e por simulações de Monte Carlo, e o segundo modelo foi investigado apenas através de simulações numéricas. O estudo do modelo carregado foi precedido por uma investigação criteriosa das técnicas de simulação de sistemas com interação Coulombiana, resultando no desenvolvimento de uma metodologia adequada a condições de contorno não isotrópicas e com custo computacional reduzido. Os modelos estatísticos propostos por nós levaram a dois resultados importantes. O modelo para camadas lipídicas neutras é, até hoje, o único modelo estatístico que descreve tanto a transição gel-fluido de bicamadas lipídicas, como a transição líquido condensado - líquido expandido\" de monocamadas de Langmuir, além de descrever também a transição líquido expandido gás na interface água-ar. O modelo para camadas lipídicas que se dissociam em água reproduz a variação abrupta na dissociação, concomitante com a transição ordem-desordem, propriedade que permite interpretar estudos experimentais relativos à condutividade das soluções lipídicas correspondentes. / Lipids in aqueous solution form a variety of different structures which include monolayers of surfactants at the water-air interface, known as Langmuir monolayers, as well as unilamellar or plurilamellar vesicles within the solution. Under temperature variation, these structures display different phases, observed through calorimetry or isothermal variation of lateral pressure. Among the phases presented by these structures, the two most important differ in the order of the lipid chains. From the point of view of the thermodynamic phases, our understanding is that Langmuir monolayers and lipid bilayers constitute the same physical system under different constraints. In this work, we develop a statistical model for the order - disorder transition of lipid bilayers which adds density fluctuations to Doniach\'s 1980 model, which has been considered the basis for many theoretical studies for lipid systems phase transitions. Density fluctuations are essential in the description of the properties of charged vesicles in solution, which consist of surfactants whose polar head dissociates in water. The study in the laboratory of thermal and structural properties of artificial charged lipid membranes is relatively new, and was developed largely in the IFUSP Laboratory of Biophysics. Such membranes exhibit distinct behavior if compared to neutral membranes, notoriously influenced by the solution salt concentration. The experimental investigations motivated us to develop a second model, in which we describe the polar headgroups through a pair of opposite charges. The lipid layer is attached to the lattice restricted primitive model, which plays the role of the saline solution. The first model was studied both through a mean-field approximation as well as through Monte Carlo simulations, whereas the second model was investigated only through numerical simulations. The study of the charged model was preceded by a thorough investigation of the simulation techniques for Coulomb interaction systems, leading to the development of a methodology suitable for non isotropic boundary conditions and with reduced computational cost. The statistical models proposed by us led to two important results. To our knowledge, our model for neutral lipid layers is the only statistical model which, aside from describing simultaneously both the gel-fluid transition of lipid bilayers and the condensed liquid - expanded liquid transition of Langmuir monolayers, also describes the gas- expanded liquid transition at the air-water interface. The model for lipid layers that dissociate in water reproduces the abrupt change in dissociation, concomitant with the order-disorder transition, a property that allows us to interpret experimental studies related to conductivity of the corresponding lipid solutions.