Tesis sobre el tema "Inférence Dynamique"
Crea una cita precisa en los estilos APA, MLA, Chicago, Harvard y otros
Consulte los 44 mejores tesis para su investigación sobre el tema "Inférence Dynamique".
Junto a cada fuente en la lista de referencias hay un botón "Agregar a la bibliografía". Pulsa este botón, y generaremos automáticamente la referencia bibliográfica para la obra elegida en el estilo de cita que necesites: APA, MLA, Harvard, Vancouver, Chicago, etc.
También puede descargar el texto completo de la publicación académica en formato pdf y leer en línea su resumen siempre que esté disponible en los metadatos.
Explore tesis sobre una amplia variedad de disciplinas y organice su bibliografía correctamente.
Delattre, Maud. "Inférence statistique dans les modèles mixtes à dynamique Markovienne". Phd thesis, Université Paris Sud - Paris XI, 2012. http://tel.archives-ouvertes.fr/tel-00765708.
Texto completoBilloir, Élise. "Modélisation dynamique et inférence bayésienne pour l'analyse de données en écotoxicologie". Lyon 1, 2008. http://www.theses.fr/2008LYO10146.
Texto completoCornuéjols, Antoine. "De l'apprentissage incrémental par adaptation dynamique : le système Influence". Paris 11, 1989. http://www.theses.fr/1989PA112003.
Texto completoFischer, Fabian. "Inférence de la structure et dynamique des forêts tropicales humides avec un modèle individu-centré". Thesis, Toulouse 3, 2019. http://www.theses.fr/2019TOU30229.
Texto completoClimate change presents society and science with a challenge that goes beyond the temporal and spatial scales of most practical problems. It therefore requires approaches that reflect the complexity of the Earth's system. This holds particularly true for the biosphere and forest ecosystems, one of the most important sources of uncertainty in climate projections. Concerted data collection efforts, such as forest inventories, trait data bases, and new technologies, such as remote sensing, have considerably increased our ability to observe and analyze the current state of the Earth's vegetation. However, to extrapolate findings into the future and understand the feedbacks between vegetation and climate change, models are needed that assimilate these data and translate them into ecosystem dynamics. Mechanistic and individual-based forest models are a particular promising approach, since they simulate dynamics bottom-up, reconstruct forests tree by tree, and are thus able to predict patterns across scales. This PhD further develops the trait- and individual-based forest growth simulator TROLL, including intraspecific variation and plasticity in tree growth, derives a new method to translate Airborne Lidar data into virtual forest inventories and uses it to infer forest structure and ecosystem dynamics in tropical rain forests. Finally, in line with TROLL's trait-based approach, an update to a global trait base, the Global Wood Density Database is presented, exploring the contribution of evolutionary lineages to wood density variation and mapping wood density across the globe
Sylvand, Benjamin. "Concept et changement de concept : Concept, contenu et inférence, Bases pour une approche dynamique du concept". Phd thesis, Université Paris-Sorbonne - Paris IV, 2005. http://tel.archives-ouvertes.fr/ijn_00000655.
Texto completoLeroux, Romain. "Inférence bayésienne pour la reconstruction d'écoulements complexes - Application au profil NACA0012". Phd thesis, Université de Poitiers, 2012. http://tel.archives-ouvertes.fr/tel-00766239.
Texto completoMunoz, François Julien. "Distribution régionale des espèces et dynamique des métapopulations : modèle hiérarchique d'habitat et inférence du taux relatif extension/colonisation". Montpellier 2, 2006. http://www.theses.fr/2006MON20012.
Texto completoA species cannot survive locally unless its biological requirements are met (niche concept). However, because of the stochasticity of its own dynamics and of the dynamics of its environment, every population is doomed to go extinct. Hence the fate of a metapopulation depends on the balance between colonization and extinction of individual populations. The floristic atlas of the French Drôme district by Luc Garraud is the basis and the motivation of our research on this topic. We consider a species' distribution as the spatial map of a single metapopulation. A global theoretical investigation of the processes involved allows us to propose new developments. We show that self-organized spatial and temporal structures are of importance. We also demonstrate that appropriate spatial statistics using spectral analysis allow to evidence metapopulation dynamics. Finally we propose an inference framework that sequentially estimates niche properties and metapopulation features. We use this framework to establish some general ecological features of plant dynamics in the Drôme district. We highlight some principles that are of importance to infer ecological processes from spatial occurrence data. The uncertainty principle means that less precise indexes of spatial structure can provide more relevant ecological information, because they filter out local contingent structures. Also, local processes should be inferred using observations at an intermediate scale and not at the scale of the overall system: this allows taking into account the effect of emergent structures. The niche concept and its spatial counterpart, the habitat, are at the meeting point of such ideas. The perspectives we propose in our work offer interesting and promising milestones in the fields of population and community ecology
Zhang, Erliang. "Etude de problèmes inverses en dynamique des structures par inférence bayésienne : recalage de modèle et reconstruction des efforts". Compiègne, 2010. http://www.theses.fr/2010COMP1857.
Texto completoThis work deals with inverse problems in structural dynamics, model updating and force reconstruction, both of them are two fundamental issues to study the dynamic behavior of a structure. Faced to the problems of non-identifiability and of calculation time, model updating consists in identifying the parameters of Finite element (FE) model. Force reconstruction is performed based on a model which is often uncertain. Due to noisy and finite expérimental data, both kinds of inverse problems are potentially ill-posed in the sense of Hadamard. In order to deal with this, a Bayesian approach domain is proposed in the frequency. FE model updating is carried out within Bayesian framework thanks to the following points. A special strategy of measurement is adopted to identify the best linear approximation of a structure and the associated standard deviation using a multi-sine excitation. The evolutionary MCMC algorithm is applied to explore the posterior probability distribution which is implicit function of the parameters. A stochastic FE model is constructed using the polynomial chaos. The FE modeling error is taken into account in the model updating process by introducing additional variables. The second part of the work uses Bayesian inference to reconstruct the force of smooth form in the frequency domain. The force is reconstructed with the Gibbs sampler, where the force is parameterized by a 1D Hermit element and segment regularization is formed. In the case of uncertain model, a conjoint approach is proposed to reconstruct the force and adjust the uncertain model. The established approach is validated by a laboratory structure
Gajda, Dorota. "Optimisation des méthodes algorithmiques en inférence bayésienne. Modélisation dynamique de la transmission d'une infection au sein d'une population hétérogène". Phd thesis, Université Paris Sud - Paris XI, 2011. http://tel.archives-ouvertes.fr/tel-00659618.
Texto completoRoncen, Rémi. "Modélisation et identification par inférence bayésienne de matériaux poreux acoustiques en aéronautique". Thesis, Toulouse, ISAE, 2018. http://www.theses.fr/2018ESAE0023/document.
Texto completoThe present work focuses on porous materials in aeronautics and the uncertainty considerations on the performed identifications. Porous materials are added inside the cavities of acoustic liners, materials formed with perforated plates and cavities, behaving as Helmholtz resonators, which are widely used in the industry. The aim is to increase the frequency range of the absorption spectrum, while improving the behaviour of liners to grazing flow and high sound intensity.This general topic is addressed by following two different leads.Porous materials were first considered in order to identify the intrinsic properties of their micro-geometry, necessary to the equivalent fluid semi-phenomenological models used later on. To achieve this, a statistical Bayesian inference tool is used to extract information on these properties, contained in reflected or transmitted signals, in three distinct frequency regimes. Furthermore, a modelling extension of rigid porous media is introduced, by adding two new intrinsic parameters related to the pore micro-structure and linked to the visco-inertial behaviour of the intra-pore fluid, at low frequencies.Then, the liner impedance, a global property representing the acoustic behaviour of materials, is identified through a Bayesian inference process. Data from a NASA benchmark are used to validate the developed tool, when the liner is subject to a shear grazing flow. An extension of these results to ONERA's B2A aeroacoustic bench is also performed, with measurements of the velocity profiles above the liner, obtained with a Laser Doppler Velocimetry technique. This identification technique is then further used for liner materials filled with porous media, to highlight the eventual influence of such a porous media on the acoustic response of the liner, when subject to a shear grazing flow. Additional measurements are permed without flow, at normal incidence, in a classical impedance tube. Different combinations of perforated plates and porous materials are tested at different sound pressure level, to evaluate the influence of the presence of porous media on the non-linear behaviour of liners when high sound pressure levels are present
Reype, Christophe. "Modélisation probabiliste et inférence bayésienne pour l’analyse de la dynamique des mélanges de fluides géologiques : détection des structures et estimation des paramètres". Electronic Thesis or Diss., Université de Lorraine, 2022. http://www.theses.fr/2022LORR0235.
Texto completoThe analysis of hydrogeochemical data aims to improve the understanding of mass transfer in the sub-surface and the Earth’s crust. This work focuses on the study of fluid-fluid interactions through fluid mixing systems, and more particularly on the detection of the compositions of the mixing sources. The detection is done by means of a point process: the proposed model is unsupervised and applicable to multidimensional data. Physical knowledge of the mixtures and geological knowledge of the data are directly integrated into the probability density of a Gibbs point process, which distributes point patterns in the data space, called the HUG model. The detected sources form the point pattern that maximises the probability density of the HUG model. This probability density is known up to the normalization constant. The knowledge related to the parameters of the model, either acquired experimentally or by using inference methods, is integrated in the method under the form of prior distributions. The configuration of the sources is obtained by a simulated annealing algorithm and Markov Chain Monte Carlo (MCMC) methods. The parameters of the model are estimated by an approximate Bayesian computation method (ABC). First, the model is applied to synthetic data, and then to real data. The parameters of the model are then estimated for a synthetic data set with known sources. Finally, the sensitivity of the model to data uncertainties, to parameters choices and to algorithms set-up is studied
Ciuciu, Philippe. "Dynamique cérébrale en neuroimagerie fonctionnelle". Habilitation à diriger des recherches, Université Paris Sud - Paris XI, 2008. http://tel.archives-ouvertes.fr/tel-00333734.
Texto completoCe thème de recherche embrasse à la fois des problèmes bas niveau relatifs à la reconstruction d'images en IRM mais aussi des aspects plus haut niveau qui concernent l'estimation et la sélection de modèles hémodynamiques régionaux non-paramétriques, capables de prendre en compte la variabilité inter-individuelle de la réponse impulsionnelle du système neuro-vasculaire. Les problèmes de reconstruction sont traités à l'aide de méthodes classiques de régularisation dans l'espace image ou des méthodes plus évoluées opérant dans l'espace transformé des coefficients d'ondelette. Les aspects inférentiels haut niveau sont majoritairement abordés dans le cadre des statistiques bayésiennes.
Bennaceur, Amel. "Synthèse dynamique de médiateurs dans les environnements ubiquitaires". Phd thesis, Université Pierre et Marie Curie - Paris VI, 2013. http://tel.archives-ouvertes.fr/tel-00849402.
Texto completoBirolleau, Alexandre. "Résolution de problème inverse et propagation d'incertitudes : application à la dynamique des gaz compressibles". Phd thesis, Université Pierre et Marie Curie - Paris VI, 2014. http://tel.archives-ouvertes.fr/tel-01023856.
Texto completoPasanisi, Alberto. "Aide à la décision dans la gestion des parcs de compteurs d'eau potable". Phd thesis, ENGREF (AgroParisTech), 2004. http://pastel.archives-ouvertes.fr/pastel-00000935.
Texto completoRozas, Rony. "Intégration du retour d'expérience pour une stratégie de maintenance dynamique". Thesis, Paris Est, 2014. http://www.theses.fr/2014PEST1112/document.
Texto completoThe optimization of maintenance strategies is a major issue for many industrial applications. It involves establishing a maintenance plan that ensures security levels, security and high reliability with minimal cost and respecting any constraints. The increasing number of works on optimization of maintenance parameters in particular in scheduling preventive maintenance action underlines the importance of this issue. A large number of studies on maintenance are based on a modeling of the degradation of the system studied. Probabilistic Models Graphics (PGM) and especially Markovian PGM (M-PGM) provide a framework for modeling complex stochastic processes. The issue with this approach is that the quality of the results is dependent on the model. More system parameters considered may change over time. This change is usually the result of a change of supplier for replacement parts or a change in operating parameters. This thesis deals with the issue of dynamic adaptation of a maintenance strategy, with a system whose parameters change. The proposed methodology is based on change detection algorithms in a stream of sequential data and a new method for probabilistic inference specific to the dynamic Bayesian networks. Furthermore, the algorithms proposed in this thesis are implemented in the framework of a research project with Bombardier Transportation. The study focuses on the maintenance of the access system of a new automotive designed to operate on the rail network in Ile-de-France. The overall objective is to ensure a high level of safety and reliability during train operation
Petiet, Florence. "Réseau bayésien dynamique hybride : application à la modélisation de la fiabilité de systèmes à espaces d'états discrets". Thesis, Paris Est, 2019. http://www.theses.fr/2019PESC2014/document.
Texto completoReliability analysis is an integral part of system design and operation, especially for systems running critical applications. Recent works have shown the interest of using Bayesian Networks in the field of reliability, for modeling the degradation of a system. The Graphical Duration Models are a specific case of Bayesian Networks, which make it possible to overcome the Markovian property of dynamic Bayesian Networks. They adapt to systems whose sojourn-time in each state is not necessarily exponentially distributed, which is the case for most industrial applications. Previous works, however, have shown limitations in these models in terms of storage capacity and computing time, due to the discrete nature of the sojourn time variable. A solution might be to allow the sojourn time variable to be continuous. According to expert opinion, sojourn time variables follow a Weibull distribution in many systems. The goal of this thesis is to integrate sojour time variables following a Weibull distribution in a Graphical Duration Model by proposing a new approach. After a presentation of the Bayesian networks, and more particularly graphical duration models, and their limitations, this report focus on presenting the new model allowing the modeling of the degradation process. This new model is called Weibull Hybrid Graphical Duration Model. An original algorithm allowing inference in such a network has been deployed. Various so built databases allowed to learn on one hand a Graphical Duration Model, and on an other hand a Graphical Duration Model Hybrid - Weibull, in order to compare them, in term of learning quality, of inference quality, of compute time, and of storage space
Lohier, Théophile. "Analyse temporelle de la dynamique de communautés végétales à l'aide de modèles individus-centrés". Thesis, Clermont-Ferrand 2, 2016. http://www.theses.fr/2016CLF22683/document.
Texto completoPlant communities are complex systems in which multiple species differing by their functional attributes interact with their environment and with each other. Because of the number and the diversity of these interactions the mechanisms that drive the dynamics of theses communities are still poorly understood. Modelling approaches enable to link in a mechanistic fashion the process driving individual plant or population dynamics to the resulting community dynamics. This PhD thesis aims at developing such approaches and to use them to investigate the mechanisms underlying community dynamics. We therefore developed two modelling approaches. The first one is based on a stochastic modelling framework allowing to link the population dynamics to the community dynamics whilst taking account of intra- and interspecific interactions as well as environmental and demographic variations. This approach is easily applicable to real systems and enables to describe the properties of plant population through a small number of demographic parameters. However our work suggests that there is no simple relationship between these parameters and plant functional traits, while they are known to drive their response to extrinsic factors. The second approach has been developed to overcome this limitation and rely on the individual-based model Nemossos that explicitly describes the link between plant functioning and community dynamics. In order to ensure that Nemossos has a large application potential, a strong emphasis has been placed on the tradeoff between realism and parametrization cost. Nemossos has then been successfully parameterized from trait values found in the literature, its realism has been demonstrated and it has been used to investigate the importance of temporal environmental variability for the coexistence of functionally differing species. The complementarity of the two approaches allows us to explore various fundamental questions of community ecology including the impact of competitive interactions on community dynamics, the effect of environmental filtering on their functional composition, or the mechanisms favoring the coexistence of plant species. In this work, the two approaches have been used separately but their coupling might offer interesting perspectives such as the investigation of the relationships between plant functioning and population dynamics. Moreover each of the approaches might be used to run various simulation experiments likely to improve our understanding of mechanisms underlying community dynamics
Frusque, Gaëtan. "Inférence et décomposition modale de réseaux dynamiques en neurosciences". Thesis, Lyon, 2020. http://www.theses.fr/2020LYSEN080.
Texto completoDynamic graphs make it possible to understand the evolution of complex systems evolving over time. This type of graph has recently received considerable attention. However, there is no consensus on how to infer and study these graphs. In this thesis, we propose specific methods for dynamical graph analysis. A dynamical graph can be seen as a succession of complete graphs sharing the same nodes, but with the weights associated with each link changing over time. The proposed methods can have applications in neuroscience or in the study of social networks such as Twitter and Facebook for example. The issue of this thesis is epilepsy, one of the most common neurological diseases in the world affecting around 1% of the population.The first part concerns the inference of dynamical graph from neurophysiological signals. To assess the similarity between each pairs of signals, in order to make the graph, we use measures of functional connectivity. The comparison of these measurements is therefore of great interest to understand the characteristics of the resulting graphs. We then compare functional connectivity measurements involving the instantaneous phase and amplitude of the signals. We are particularly interested in a measure called Phase-Locking-Value (PLV) which quantifies the phase synchrony between two signals. We then propose, in order to infer robust and interpretable dynamic graphs, two new indexes that are conditioned and regularized PLV. The second part concerns tools for dynamical graphs decompositions. The objective is to propose a semi-automatic method in order to characterize the most important patterns in the pathological network from several seizures of the same patient. First, we consider seizures that have similar durations and temporal evolutions. In this case the data can be conveniently represented as a tensor. A specific tensor decomposition is then applied. Secondly, we consider seizures that have heterogeneous durations. Several strategies are proposed and compared. These are methods which, in addition to extracting the characteristic subgraphs common to all the seizures, make it possible to observe their temporal activation profiles specific to each seizures. Finally, the selected method is used for a clinical application. The obtained decompositions are compared to the visual interpretation of the clinician. As a whole, we found that activated subgraphs corresponded to brain regions involved during the course of the seizures and their time course were highly consistent with classical visual interpretation
Guggiola, Alberto. "Une approche physique-statistique à différents problèmes dans la théorie des réseaux". Thesis, Paris, Ecole normale supérieure, 2015. http://www.theses.fr/2015ENSU0034/document.
Texto completoStatistical physics, originally developed to describe thermodynamic systems, has been playing for the last decades a central role in modelling an incredibly large and heterogeneous set of different phenomena taking for instance place on social, economical or biological systems. Such a vast field of possible applications has been found also for networks, as a huge variety of systems can be described in terms of interconnected elements. After an introductory part introducing these themes as well as the role of abstract modelling in science, in this dissertation it will be discussed how a statistical physics approach can lead to new insights as regards three problems of interest in network theory: how some quantity can be optimally spread on a graph, how to explore it and how to reconstruct it from partial information. Some final remarks on the importance such themes will likely preserve in the coming years conclude the work
Guedj, Jérémie. "Inférence dans les modèles dynamiques de population : applications au VIH et au VHC". Bordeaux 2, 2006. http://www.theses.fr/2006BOR21371.
Texto completoThe study of the dynamical models of HIV, based on non- linear systems of Ordinary Differential Equations has considerably improved the knowledge on its pathogenicity. This modelling leads to complex issues for identifiability and parameter estimation. To overcome these difficulties, the first models used simplified ODE systems and analyzed each patient separately. However, these models prevent from considering the course of the infection as a whole. We propose here an alternative way based on a full likelihood inference, using the particular structure of the non-simplified models and borrowing strength from the whole sample. We apply it to real data, taking into account the viral load left-censoring, and we illustrate the interest of this approach to provide an alternative tool for analyzing clinical trials. Last, we study the practical identifiability of these models
Bonnaffoux, Arnaud. "Inférence de réseaux de régulation de gènes à partir de données dynamiques multi-échelles". Thesis, Lyon, 2018. http://www.theses.fr/2018LYSEN054/document.
Texto completoInference of gene regulatory networks from gene expression data has been a long-standing and notoriously difficult task in systems biology. Recently, single-cell transcriptomic data have been massively used for gene regulatory network inference, with both successes and limitations.In the present work we propose an iterative algorithm called WASABI, dedicated to inferring a causal dynamical network from timestamped single-cell data, which tackles some of the limitations associated with current approaches. We first introduce the concept of waves, which posits that the information provided by an external stimulus will affect genes one-byone through a cascade, like waves spreading through a network. This concept allows us to infer the network one gene at a time, after genes have been ordered regarding their time of regulation. We then demonstrate the ability of WASABI to correctly infer small networks, which have been simulated in-silico using a mechanistic model consisting of coupled piecewise-deterministic Markov processes for the proper description of gene expression at the single-cell level. We finally apply WASABI on in-vitro generated data on an avian model of erythroid differentiation. The structure of the resulting gene regulatory network sheds a fascinating new light on the molecular mechanisms controlling this process. In particular, we find no evidence for hub genes and a much more distributed network structure than expected. Interestingly, we find that a majority of genes are under the direct control of the differentiation-inducing stimulus. Together, these results demonstrate WASABI versatility and ability to tackle some general gene regulatory networks inference issues. It is our hope that WASABI will prove useful in helping biologists to fully exploit the power of time-stamped single-cell data
Lèbre, Sophie. "Analyse de processus stochastiques pour la génomique : étude du modèle MTD et inférence de réseaux bayésiens dynamiques". Evry-Val d'Essonne, 2007. http://www.biblio.univ-evry.fr/theses/2007/interne/2007EVRY0017.pdf.
Texto completoThis thesis deals with DNA sequence and time series gene expression analysis. First we study the parsimonious Markov model called Mixture Transition Distribution (MTD) model and introduce an EM algorithm for MTD models estimation. Then we propose two approaches for genetic network recovering using Dynamic Bayesian Networks (DBNs). The dependencies are described by a directed graph whose topology has to be inferred despite the overly low number of repeated measurements compared with the number of observed genes. First we assume that the topology is constant across time, we approximate this graph by considering partial order dependencies and we develop a deterministic procedure for DBNs inference. Then we consider a multiple changepoint regression model defining a succession of homogeneous phases. The changepoints location and the structure within each phase are simultaneously inferred thanks to a reversible jump MCMC procedure
Bouzidi, Halima. "Efficient Deployment of Deep Neural Networks on Hardware Devices for Edge AI". Electronic Thesis or Diss., Valenciennes, Université Polytechnique Hauts-de-France, 2024. http://www.theses.fr/2024UPHF0006.
Texto completoNeural Networks (NN) have become a leading force in today's digital landscape. Inspired by the human brain, their intricate design allows them to recognize patterns, make informed decisions, and even predict forthcoming scenarios with impressive accuracy. NN are widely deployed in Internet of Things (IoT) systems, further elevating interconnected devices' capabilities by empowering them to learn and auto-adapt in real-time contexts. However, the proliferation of data produced by IoT sensors makes it difficult to send them to a centralized cloud for processing. This is where the allure of edge computing becomes captivating. Processing data closer to where it originates -at the edge- reduces latency, makes real-time decisions with less effort, and efficiently manages network congestion.Integrating NN on edge devices for IoT systems enables more efficient and responsive solutions, ushering in a new age of self-sustaining Edge AI. However, Deploying NN on resource-constrained edge devices presents a myriad of challenges: (i) The inherent complexity of neural network architectures, which requires significant computational and memory capabilities. (ii) The limited power budget of IoT devices makes the NN inference prone to rapid energy depletion, drastically reducing system utility. (iii) The hurdle of ensuring harmony between NN and HW designs as they evolve at different rates. (iv) The lack of adaptability to the dynamic runtime environment and the intricacies of input data.Addressing these challenges, this thesis aims to establish innovative methods that extend conventional NN design frameworks, notably Neural Architecture Search (NAS). By integrating HW and runtime contextual features, our methods aspire to enhance NN performances while abstracting the need for the human-in-loop}. Firstly, we incorporate HW properties into the NAS by tailoring the design of NN to clock frequency variations (DVFS) to minimize energy footprint. Secondly, we leverage dynamicity within NN from a design perspective, culminating in a comprehensive Hardware-aware Dynamic NAS with DVFS features. Thirdly, we explore the potential of Graph Neural Networks (GNN) at the edge by developing a novel HW-aware NAS with distributed computing features on heterogeneous MPSoC. Fourthly, we address the SW/HW co-optimization on heterogeneous MPSoCs by proposing an innovative scheduling strategy that leverages NN adaptability and parallelism across computing units. Fifthly, we explore the prospect of ML4ML -- Machine Learning for Machine Learning by introducing techniques to estimate NN performances on edge devices using neural architectural features and ML-based predictors. Finally, we develop an end-to-end self-adaptive evolutionary HW-aware NAS framework that progressively learns the importance of NN parameters to guide the search process toward Pareto optimality effectively.Our methods can contribute to elaborating an end-to-end design framework for neural networks on edge hardware devices. They enable leveraging multiple optimization opportunities at both the software and hardware levels, thus improving the performance and efficiency of Edge AI
Boualem, Abdelbassit. "Estimation de distribution de tailles de particules par techniques d'inférence bayésienne". Thesis, Orléans, 2016. http://www.theses.fr/2016ORLE2030/document.
Texto completoThis research work treats the inverse problem of particle size distribution (PSD) estimation from dynamic light scattering (DLS) data. The current DLS data analysis methods have bad estimation results repeatability and poor ability to separate the components (resolution) of a multimodal sample of particles. This thesis aims to develop new and more efficient estimation methods based on Bayesian inference techniques by taking advantage of the angular diversity of the DLS data. First, we proposed a non-parametric method based on a free-form model with the disadvantage of requiring a priori knowledge of the PSD support. To avoid this problem, we then proposed a parametric method based on modelling the PSD using a Gaussian mixture model. The two proposed Bayesian methods use Markov chain Monte Carlo simulation algorithms. The obtained results, on simulated and real DLS data, show the capability of the proposed methods to estimate multimodal PSDs with high resolution and better repeatability. We also computed the Cramér-Rao bounds of the Gaussian mixture model. The results show that there are preferred angle values ensuring minimum error on the PSD estimation
Prieur, Clémentine. "Dépendance faible: estimation et théorèmes limite.Application à l'étude statistique de certains systèmes dynamiques". Habilitation à diriger des recherches, Université Paul Sabatier - Toulouse III, 2006. http://tel.archives-ouvertes.fr/tel-00133468.
Texto completonon -mélangeantes au sens de Rosenblatt (1956). La notion de mélange classique est affaiblie
afin d'établir des inégalités ainsi que des théorèmes limite pour différentes classes de processus
comme par exemple certains systèmes dynamiques, des chaînes de Markov non irréductibles,
ou encore des fonctions de processus linéaires non mélangeants. Les résultats obtenus sont
ensuite appliqués au domaine de la statistique non paramétrique.
Deux autres thématiques sont abordées dans ce manuscrit : d'une part l'étude de principes
de grandes déviations (notamment pour le processus de records généralisés), et d'autre part
l'estimation adaptative de fonctionnelles linéaires.
Castaño, Maria Soledad. "Hétérogénéité dans des processus de développement cachés : inférence et analyse de populations structurées en environnements fluctuants". Thesis, Antilles, 2017. http://www.theses.fr/2017ANTI0124/document.
Texto completoCodakia orbicularis is a bivalve mollusk belonging to the family Lucinidae harboring sulfur-oxidizing bacterial endosymbionts within its gills. Considering that any symbiosis is most likely regulated by dialogue molecules, an exhaustive chemical study could lead to identify the involved metabolites. Thus, the aim of this thesis focuses on the isolation of secondary metabolites from the gills of this bivalve and the evaluation of the antibacterial activity of the isolated molecules. Twelve compounds were isolated from the gills of Codakia orbicularis and their structures were determined by usual spectroscopic methods. Among these molecules, only one presented a new structure and has been named orbicularisine. The latter presents an undescribed spirotetracyclic indolone skeleton. Regarding the biological activities, among the isolated molecules, only four of them identified as S8 sulfur, 4-hydroxybenzaldehyde and two monoglycerolipids presented an antibacterial activity. Orbicularisine was inactive against a panel of cell lines and kinase. The orbicularisine new skeleton is an interesting start for the synthesis of new family of molecules, thus enhancing its molecular diversity. It will be interesting to determine the origin of the isolated molecules (prokaryotic or eukaryotic), especially for the new orbicularisine, and their roles in the frame of the symbiosis. The chemical results obtained on C. orbicularis and on lucinids in general are interesting since the coastal species belonging to Bivalves have not been chemically explored
Chevalier, Stéphanie. "Inférence logique de réseaux booléens à partir de connaissances et d'observations de processus de différenciation cellulaire". Electronic Thesis or Diss., université Paris-Saclay, 2022. http://www.theses.fr/2022UPASG061.
Texto completoDynamic models are essential tools for exploring regulatory mechanisms in biology. This thesis was guided by the need expressed in oncology and developmental biology to automatically infer Boolean networks reproducing cellular differentiation processes.By considering observations and knowledge that the modelers have at their disposal, this thesis presents an approach that allows to model the richness of this cellular behavior by inferring all the compatible Boolean networks at that scale of the regulatory networks commonly considered in biology.To develop this method, three main contributions are presented.The first contribution is a formal framework of the properties of data collected to study cellular differentiation. This framework allows reasoning about the desired dynamic properties within Boolean networks to be consistent with this cellular behavior.The second contribution concerns the encoding of the model inference problem as a Boolean satisfiability problem whose solutions are the Boolean networks compatible with the biological data. For this, constraints on the dynamics of Boolean networks corresponding to the previously formalized properties have been implemented in logic programming.The last contribution was to apply to real biological problems the model inference method, named BoNesis, which was developed thanks to the constraints. These applications showed the benefit of inferring a set of models for the process analysis and illustrated the modeling methodology, from the preparation of biological data to the analysis of the inferred models
Leurent, Fabien. "Modélisation du trafic, des déplacements sur un réseau et de l'accessibilité aux activités grâce au transport". Habilitation à diriger des recherches, Université Paris Dauphine - Paris IX, 2006. http://tel.archives-ouvertes.fr/tel-00348286.
Texto completoUne telle modélisation comporte quatre aspects : un contenu sémantique, à caractère physique ou économique ; une formulation mathématique ; un solveur technique ; un aspect empirique (métrologie, statistique, économétrie).
Les disciplines mises en œuvre sont variées : théorie des réseaux, optimisation, informatique algorithmique, probabilités et statistiques, et aussi économie, socio-économie et physique du trafic. Mes contributions théoriques concernent la théorie des réseaux, l'économie du transport et la physique du trafic.
Mes travaux se répartissent en quatre thèmes :
A. La mesure et la modélisation du trafic. Au niveau local d'une route, j'ai analysé la relation entre flux et vitesse en mettant en cohérence l'analyse désagrégée, probabiliste au niveau du mobile individuel ; et l'analyse macroscopique en termes de flux et de distribution statistique des temps.
B. La modélisation des réseaux et des cheminements. L'équilibre entre offre de transport et demande de déplacement conjugue une dimension spatiale - topologique, une dimension temporelle, et une dimension comportementale - économique. Les enjeux de modélisation concernent : la représentation de l'offre et la demande ; la formulation et les propriétés d'existence – unicité – stabilité ; les algorithmes. Je me suis intéressé à la diversité des comportements ; et à la modélisation fine de l'offre et à la dimension temporelle.
C. L'analyse socio-économique des déplacements. Je me suis intéressé à l'usage de divers moyens de transport et à la prospection de leur clientèle potentielle ; au choix d'horaire de déplacement ; aux caractéristiques à la fois économiques et dynamiques de la congestion.
D. La distribution spatiale des déplacements et des activités. Je me suis intéressé d'une part à l'observation des flux par relation origine-destination (O-D) et à l'inférence statistique des matrices O-D ; et d'autre part, à la justification microéconomique des déplacements en raison de la localisation et de l'utilité des activités.
Landry, Clara. "Modélisation des dynamiques de maladies foliaires de cultures pérennes tropicales à différentes échelles spatiales : cas de la cercosporiose noire du bananier". Thesis, Antilles-Guyane, 2015. http://www.theses.fr/2015AGUY0835/document.
Texto completoThis thesis concerns the modeling of the dynamics of foliar diseases of tropical tree crops at different spatial scales. This modeling approach is applied to black Sigatoka of banana. It is to explore and determine the environmental parameters and host resistance has a significant influence on the spatiotemporal dynamics of the disease and to provide elements associated with the control of Sigatoka noire.Deux models were developed as part of this thesis. The epidemiological dynamics at the plant is described by a mechanistic model decomposed into a growth model of the plant and an epidemiological unit describing the pathogen epidemic cycle. The architecture of the banana is taken into account through foliar compartments positioned in space. The model was validated by an independent data set. Numerical experiments and model sensitivity analysis performed by the methods of Morris and e-FAST enabled to better understand the functioning of this epidemic disease and identify the parameters affecting the most dynamic epidemic especially speed extension of the lesions, the incubation period and efficiency of Bayesian .L'approche infection allowed to take into account prior information available for these three parameters that were the statistical inference. The model sensitivity analysis also identified the influence of two parameters related to plant growth: the number of leaves on a plant and rate of leaf emission A dynamic model of space-time Black Sigatoka has been developed at the scale of a territory from surveys enMartinique during the period of invasion of the disease from September 2010 to May 2012. The data collected being censored, inference of model parameters has was performed in a Bayesian framework, using a data augmentation algorithm. The model developed and inference possible to reconstruct the spatiotemporal dynamics of the invasion and predict the end of the invasion territoire.Les two spatio-temporal dynamics models developed at two different spatial scales has been gained important information for build tools design method of control of black Sigatoka of banana
Abboud, Candy. "Inférer et prédire les dynamiques d'espèces invasives : focus sur Xylella fastidiosa". Thesis, Aix-Marseille, 2019. http://www.theses.fr/2019AIXM0638.
Texto completoThe thesis research aims to provide a generic methodology that improves the predictions of an invasive species dynamics for which no dedicated model is available and whose initial conditions are unknown. In order to achieve this goal, we proceed in two complementary lines of research. The first one is to propose a model&data-based inference method of biological invasions, in the framework of the so-called mechanistic-statistical approach. This method allows us to jointly estimate the introduction point and other parameters of the dynamics related to diffusion, reproduction and death. It is hinged on (i) a partial differential equation (PDE) that offers a concise description of the invasive species dynamics in a heterogeneous domain, (ii) a stochastic model that represents the observation process and (iii) a statistical Bayesian inference procedure for estimating model parameters. We propose to replace the PDE by a model issued from the framework of Piecewise-deterministic Markov Process to balance the trade-off between model realism and estimation easiness. The second research line consists on accounting for the uncertainty about models form using the Bayesian model-averaging. This method consists of combining predictions drawn from competing models in order to obtain a unique and ameliorated prediction. This technique is not widespread in the field of epidemiology. One of the methodological goals of the PhD is to investigate its application and usefulness in predictive epidemiology. The case study of my thesis is the phytopathogenic bacterium Xylella fastidiosa which is susceptible to cause in France a major sanitary crisis as the one caused in Italy since 2013
Grigolon, Silvia. "Modelling and inference for biological systems : from auxin dynamics in plants to protein sequences". Thesis, Paris 11, 2015. http://www.theses.fr/2015PA112178/document.
Texto completoAll biological systems are made of atoms and molecules interacting in a non- trivial manner. Such non-trivial interactions induce complex behaviours allow- ing organisms to fulfill all their vital functions. These features can be found in all biological systems at different levels, from molecules and genes up to cells and tissues. In the past few decades, physicists have been paying much attention to these intriguing aspects by framing them in network approaches for which a number of theoretical methods offer many powerful ways to tackle systemic problems. At least two different ways of approaching these challenges may be considered: direct modeling methods and approaches based on inverse methods. In the context of this thesis, we made use of both methods to study three different problems occurring on three different biological scales. In the first part of the thesis, we mainly deal with the very early stages of tissue development in plants. We propose a model aimed at understanding which features drive the spontaneous collective behaviour in space and time of PINs, the transporters which pump the phytohormone auxin out of cells. In the second part of the thesis, we focus instead on the structural properties of proteins. In particular we ask how conservation of protein function across different organ- isms constrains the evolution of protein sequences and their diversity. Hereby we propose a new method to extract the sequence positions most relevant for protein function. Finally, in the third part, we study intracellular molecular networks that implement auxin signaling in plants. In this context, and using extensions of a previously published model, we examine how network structure affects network function. The comparison of different network topologies provides insights into the role of different modules and of a negative feedback loop in particular. Our introduction of the dynamical response function allows us to characterize the systemic properties of the auxin signaling when external stimuli are applied
Jaoua, Nouha. "Estimation Bayésienne non Paramétrique de Systèmes Dynamiques en Présence de Bruits Alpha-Stables". Phd thesis, Ecole Centrale de Lille, 2013. http://tel.archives-ouvertes.fr/tel-00929691.
Texto completoVacher, Jonathan. "Synthèse de textures dynamiques pour l'étude de la vision en psychophysique et électrophysiologie". Thesis, Paris Sciences et Lettres (ComUE), 2017. http://www.theses.fr/2017PSLED005/document.
Texto completoThe goal of this thesis is to propose a mathematical model of visual stimulations in order to finely analyze experimental data in psychophysics and electrophysiology. More precisely, it is necessary to develop a set of dynamic, stochastic and parametric stimulations in order to exploit data analysis techniques from Bayesian statistics and machine learning. This problem is important to understand the visual system capacity to integrate and discriminate between stimuli. In particular, the measures performed at different scales (neurons, neural population, cognition) allow to study the particular sensitivities of neurons, their functional organization and their impact on decision making. To this purpose, we propose a set of theoretical, numerical and experimental contributions organized around three principal axes: (1) a Gaussian dynamic texture synthesis model specially crafted to probe vision; (2) a Bayesian observer model that accounts for the positive effect of spatial frequency over speed perception; (3) the use of machine learning techniques to analyze voltage sensitive dye optical imaging and extracellular data. This work, at the crossroads of neurosciences, psychophysics and mathematics is the fruit of several interdisciplinary collaborations
Matulewicz, Gustaw. "Statistical inference of Ornstein-Uhlenbeck processes : generation of stochastic graphs, sparsity, applications in finance". Thesis, Université Paris-Saclay (ComUE), 2017. http://www.theses.fr/2017SACLX066/document.
Texto completoThe subject if this thesis is the statistical inference of multi-dimensional Ornstein-Uhlenbeck processes. In a first part, we introduce a model of stochastic graphs, defined as binary observations of a trajectory. We show then that it is possible to retrieve the dynamic of the underlying trajectory from the binary observations. For this, we build statistics of the stochastic graph and prove new results on their convergence in the long-time, high-frequency setting. We also analyse the properties of the stochastic graph from the point of view of evolving networks. In a second part, we work in the setting of complete information and continuous time. We add then a sparsity assumption applied to the drift matrix coefficient of the Ornstein-Uhlenbeck process. We prove sharp oracle inequalities for the Lasso estimator, construct a lower bound on the estimation error for sparse estimators and show optimality properties of the Adaptive Lasso estimator. Then, we apply the methods to estimate mean-return properties of real-world financial datasets: daily returns of SP500 components and EURO STOXX 50 Dividend Future prices
El, Beheiry Mohamed Hossam. "Towards whole-cell mapping of single-molecule dynamics". Thesis, Paris 6, 2015. http://www.theses.fr/2015PA066618/document.
Texto completoImaging of single molecules inside living cells confers insight to biological function at its most granular level. Single molecules experience a nanoscopic environment that is complicated, and in general, poorly understood. The modality of choice for probing this environment is live-cell localisation microscopy, where trajectories of single molecules can be captured. For many years, the great stumbling block in comprehension of physical processes at this scale was the lack of information accessible; statistical significance and robust assertions are hardly possible from a few dozen trajectories. It is the onset of high-density single-particle tracking that has dramatically reframed the possibilities of such studies. Importantly, the consequential amounts of data it provides invites the use of powerful statistical tools that assign probabilistic descriptions to experimental observations. In this thesis, Bayesian inference tools have been developed to elucidate the behaviour of single molecules via the mapping of motion parameters. As a readout, maps describe heterogeneities at local and whole-cell scales. Importantly, they grant quantitative details into basic cellular processes. This thesis uses the mapping approach to study receptor-scaffold interactions inside neurons and non-neuronal cells. A promising system in which interactions are patterned is also examined. It is shown that interactions of different types of chimeric glycine receptors to the gephyrin scaffold protein may be described and distinguished in situ. Finally, the prospects of whole-cell mapping in three-dimensions are evaluated based on a discussion of state-of-the-art volumetric microscopy techniques
Alari, Anna. "Variations temporelles et géographiques des méningites à pneumocoque et effet du vaccin conjugué en France". Thesis, Université Paris-Saclay (ComUE), 2018. http://www.theses.fr/2018SACLV070/document.
Texto completoStreptococcus pneumoniae is a Gram-positive commensal bacterium of the oropharyngeal flora usually colonizing human’s rhino pharynx, of which almost 100 serotypes are known. Infants and young children constitute its main reservoir. Pneumococcus may cause serious infections, such as meningitis, bacteremia and pneumonia, or less serious but more common such as sinusitis and acute otitis media (AOM). Two conjugate pneumococcal vaccines have been introduced in France: PCV7 (covering 7 serotypes) in 2003 and PCV13 (covering 6 additional serotypes) in 2010. The overall objective of this thesis is to assess the impact of vaccination policy on invasive pneumococcal diseases in France, by focusing on temporal and geographical trends of the most serious of them: pneumococcal meningitis (PM). An initial study of PMs temporal dynamics over the 2011-2014 period assessed the impact of conjugate vaccines’ introduction. Statistical modeling techniques were used for time series analysis. The results confirm the effects found in literature: a reduction of vaccine serotypes PMs but at the same time an increase of PMs, due to non-vaccine serotypes (effect of “serotype replacement”). Therefore, the first benefit of vaccine introduction at population scale has been observed no less than 11 years after PCV7 introduction, and then principally after PCV13 was introduced in 2010, with a 25% decrease in PMs in 2014. The geographic component was then implemented to analyze the role of vaccine coverage in annual PM variability between geographic units over the 2001-2016 period. Results confirm the effectiveness of both vaccine compositions on vaccine serotypes PMs and suggest homogeneity of this effect among geographic units. Conversely the serotype replacement has been confirmed only after the first vaccine composition was introduced and presents a variable and heterogeneous geographical repartition. Variability in vaccine coverage among geographic units doesn’t explain the differences in PMs, which could suggest the role of others factors such as demographic density. Finally, a dynamic modeling capable of taking into consideration fundamental aspects of pneumococcus transmission and infection mechanisms not integrated in static modeling has been proposed in order to predict the impacts of different vaccination strategies for 65+ adults and therefore assess their cost-utility ratios
Nguyen, Van Duong. "Variational deep learning for time series modelling and analysis : applications to dynamical system identification and maritime traffic anomaly detection". Thesis, Ecole nationale supérieure Mines-Télécom Atlantique Bretagne Pays de la Loire, 2020. http://www.theses.fr/2020IMTA0227.
Texto completoThis thesis work focuses on a class of unsupervised, probabilistic deep learning methods that use variational inference to create high capacity, scalable models for time series modelling and analysis. We present two classes of variational deep learning, then apply them to two specific problems related to the maritime domain. The first application is the identification of dynamical systems from noisy and partially observed data. We introduce a framework that merges classical data assimilation and modern deep learning to retrieve the differential equations that control the dynamics of the system. Using a state space formulation, the proposed framework embeds stochastic components to account for stochastic variabilities, model errors and reconstruction uncertainties. The second application is maritime traffic surveillance using AIS data. We propose a multitask probabilistic deep learning architecture can achieve state-of-the-art performance in different maritime traffic surveillance related tasks, such as trajectory reconstruction, vessel type identification and anomaly detection, while reducing significantly the amount data to be stored and the calculation time. For the most important task—anomaly detection, we introduce a geospatial detector that uses variational deep learning to builds a probabilistic representation of AIS trajectories, then detect anomalies by judging how likely this trajectory is
Bellot, Benoit. "Améliorer les connaissances sur les processus écologiques régissant les dynamiques de populations d'auxiliaires de culture : modélisation couplant paysages et populations pour l'aide à l'échantillonnage biologique dans l'espace et le temps". Thesis, Rennes 1, 2018. http://www.theses.fr/2018REN1B008/document.
Texto completoA promising alternative to the chemical control of pests consists in favoring their natural enemies populations by managing the agricultural landscape structure. Identifying favorable spatio-temporal structures can be performed through the exploration of landscape scenarios using coupled models of landscapes and population dynamics. In this approach, population dynamics are simulated on virtual landscapes with controlled properties, and the observation of population patterns allows for the identification of favorable structures. Population modeling however relies on a good knowledge about the ecological processes and their variability within the landscape elements. Current state of knowledge about the ecological mechanisms underlying natural enemies’ of the carabid family population dynamics remains a major obstacle to in silico investigation of favorable landscape scenarios. Literature about the relationship between carabid population and landscape properties allows the formulation of competing hypotheses about these processes. Reducing the number of these hypotheses by analyzing the convergence between their associated population patterns and investigating the stability of their convergence along a landscape gradient appears to be a necessary tep towards a better knowledge about ecological processes. In a first step, we propose a heuristic method based on the simulation of reaction-diffusion models carrying these competing hypotheses. Comparing the population patterns allowed to set a model typology according to their response to the landscape variable, through a classification algorithm, thus reducing the initial number of competing hypotheses. The selection of the most likely hypothesis from this irreducible set must rely on the observation of population patterns on the field. This implies that population patterns are described with spatial and temporal resolutions that are fine enough to select a unique hypothesis among the ones in competition. In the second part, we propose a heuristic method that allows determining a priori sampling strategies that maximize the robustness of ecological hypotheses selection. The simulation of reaction-diffusion models carrying the ecological hypotheses allows to generate virtual population data in space and time. These data are then sampled using strategies differing in the total effort, number of sampling locations, dates and landscape replicates. Population patterns are described from these samples. The sampling strategies are assessed through a classification algorithm that classifies the models according to the associated patterns. The analysis of classification performances, i.e. the ability of the algorithm to discriminate the ecological processes, allows the selection of optimal sampling designs. We also show that the way the sampling effort is distributed between its spatial and temporal components is strongly impacting the ecological processes inference. Reducing the number of competing ecological hypotheses, along with the selection of sampling strategies for optimal model inference both meet a strong need in the process of knowledge improvement about the ecological processes for the exploration of landscape scenarios favoring ecosystem services. In the last chapter, we discuss the implications and future prospects of our work
Drulhe, Samuel. "Identification de réseaux de régulation génique à partir de données d'expression : une approche basée sur les modèles affines par morceaux". Phd thesis, Université Joseph Fourier (Grenoble), 2008. http://tel.archives-ouvertes.fr/tel-00380505.
Texto completoLa méthode que nous présentons se concentrent sur le problème de la détection des transitions entre les différents modes dynamiques à partir des données d'expression génique et sur la reconstruction des seuils de transition associés avec les interactions régulatrices. En particulier, notre méthode prend en considération les contraintes géométriques spécifiques aux modèles APM de RRGs. Une telle méthode d'identification est conçue pour des systèmes à erreur sur la sortie où les observations sont des séries temporelles de mesures bruitées de niveaux de concentration à l'intérieur d'une cellule.
Les données sont d'abord classées en modes dans lesquels le comportement dynamique est considéré comme étant complètement décrit par une équation différentielle linéaire. À partir de la classification résultante, une technique de reconnaissance de forme est utilisée pour reconstruire toutes les combinaisons de seuils de transition qui sont cohérentes avec les données mesurées. Pour chaque combinaison de seuils, il est alors possible de fournir un réseau de régulation et les paramètres dynamiques de chaque mode.
Les performances de notre approche ont été analysées en utilisant des données artificielles simulées pour un modèle simplifié de la réponse à un manque de carbone pour le bactérie Escherichia coli. En particulier, nous avons évalué l'influence du niveau du bruit et du pas d'échantillonnage sur les systèmes identifiés. Nos résultats montrent que la méthode, en association avec des séries temporelles de mesures suffisamment précises, lesquelles peuvent être obtenues avec des systèmes à gène rapporteur, permettent une identification quantitative de modèles APM de RRGs.
Papamichail, Chrysanthi. "Estimation of dynamical systems with application in mechanics". Thesis, Compiègne, 2016. http://www.theses.fr/2016COMP2284.
Texto completoThe present dissertation is devoted to the statistical inference, bootstrap methods and multivariate analysis in the framework of semi-Markov processes. The main applications concern a mechanical problem from fracture mechanics. This work has a two-fold contribution. The first part concerns in general the stochastic modeling of the fatigue crack propagation phenomenon. A stochastic differential equation describes the degradation mechanism and the innate randomness of the phenomenon is handled by a perturbation process. Under the assumption that this process is a jump Markov (or semi-Markov) process, the reliability of the model is studied by means of Markov renewal theory and a new, faster, reliability calculus method is proposed with the respective algorithm. The method and the model for the Markov perturbation process are validated on experimental fatigue data. Next, the strong consistency of the least squares estimates of the model parameters is obtained by assuming that the residuals of the stochastic regression model are martingale differences into which the initial model function is transformed. In the second part of the manuscript, we have tackled the difficult problem of approximating the limiting distribution of certain non-parametric estimators of semi-Markov kernels or some functionals of them via the weighted bootstrap methodology in a general framework. Applications of these results on statistical problems such as the construction of confidence bands, the statistical tests, the computation of the p-value of the test are provided and the estimation of the generalized inverses
Deregnaucourt, Thomas. "Prédiction spatio-temporelle de surfaces issues de l'imagerie en utilisant des processus stochastiques". Thesis, Université Clermont Auvergne (2017-2020), 2019. http://www.theses.fr/2019CLFAC088.
Texto completoThe prediction of a surface is now an important problem due to its use in multiple domains, such as computer vision, the simulation of avatars for cinematography or video games, etc. Since a surface can be static or dynamic, i.e. evolving with time, this problem can be separated in two classes: a spatial prediction problem and a spatio-temporal one. In order to propose a new approach for each of these problems, this thesis works have been separated in two parts.First of all, we have searched to predict a static surface, which is supposed cylindrical, knowing it partially from curves. The proposed approach consisted in deforming a cylinder on the known curves in order to reconstruct the surface of interest. First, a correspondence between known curves and the cylinder is generated with the help of shape analysis tools. Once this step done, an interpolation of the deformation field, which is supposed Gaussian, have been estimated using maximum likelihood and Bayesian inference. This methodology has then been applied to real data from two domains of imaging: medical imaging and infography. The obtained results show that the proposed approach exceeds the existing methods in the literature, with better results using Bayesian inference.In a second hand, we have been interested in the spatio-temporal prediction of dynamic surfaces. The objective was to predict a dynamic surface based on its initial surface. Since the prediction needs to learn on known observations, we first have developed a spatio-temporal surface analysis tool. This analysis is based on shape analysis tools, and allows a better learning. Once this preliminary step done, we have estimated the temporal deformation of the dynamic surface of interest. More precisely, an adaptation, with is usable on the space of surfaces, of usual statistical estimators has been used. Using this estimated deformation on the initial surface, an estimation of the dynamic surface has been created. This process has then been applied for predicting 4D expressions of faces, which allow us to generate visually convincing expressions
Rau, Andrea. "Inférence rétrospective de réseaux de gènes à partir de données génomiques temporelles". Phd thesis, 2010. http://tel.archives-ouvertes.fr/tel-00568663.
Texto completoDans ce travail, nous proposons deux méthodes pour l'identification des réseaux de gènes régulateurs qui se servent des réseaux Bayésiens dynamiques et des modèles linéaires. Dans la première méthode, nous développons un algorithme dans un cadre bayésien pour les modèles linéaires espace-état (state-space model). Les hyperparamètres sont estimés avec une procédure bayésienne empirique et une adaptation de l'algorithme espérance-maximisation. Dans la deuxième approche, nous développons une extension d'une méthode de Approximate Bayesian Computation basé sur une procédure de Monte Carlo par chaînes de Markov pour l'inférence des réseaux biologiques. Cette méthode échantillonne des lois approximatives a posteriori des interactions gène-à-gène et fournit des informations sur l'identifiabilité et le robustesse des structures sous-réseaux. La performance des deux approches est étudié via un ensemble de simulations, et les deux sont appliqués aux données transcriptomiques.
Sankar, Chinnadhurai. "Neural approaches to dialog modeling". Thesis, 2020. http://hdl.handle.net/1866/24802.
Texto completoThis thesis by article consists of four articles which contribute to the field of deep learning, specifically in understanding and learning neural approaches to dialog systems. The first article takes a step towards understanding if commonly used neural dialog architectures effectively capture the information present in the conversation history. Through a series of perturbation experiments on popular dialog datasets, wefindthatcommonly used neural dialog architectures like recurrent and transformer-based seq2seq models are rarely sensitive to most input context perturbations such as missing or reordering utterances, shuffling words, etc. The second article introduces a simple and cost-effective way to collect large scale datasets for modeling task-oriented dialog systems. This approach avoids the requirement of a com-plex argument annotation schema. The initial release of the dataset includes 13,215 task-based dialogs comprising six domains and around 8k unique named entities, almost 8 times more than the popular MultiWOZ dataset. The third article proposes to improve response generation quality in open domain dialog systems by jointly modeling the utterances with the dialog attributes of each utterance. Dialog attributes of an utterance refer to discrete features or aspects associated with an utterance like dialog-acts, sentiment, emotion, speaker identity, speaker personality, etc. The final article introduces an embedding-free method to compute word representations on-the-fly. This approach significantly reduces the memory footprint which facilitates de-ployment in on-device (memory constraints) devices. Apart from being independent of the vocabulary size, we find this approach to be inherently resilient to common misspellings.