Дисертації з теми "Sélection des modèles"
Оформте джерело за APA, MLA, Chicago, Harvard та іншими стилями
Ознайомтеся з топ-50 дисертацій для дослідження на тему "Sélection des modèles".
Біля кожної праці в переліку літератури доступна кнопка «Додати до бібліографії». Скористайтеся нею – і ми автоматично оформимо бібліографічне посилання на обрану працю в потрібному вам стилі цитування: APA, MLA, «Гарвард», «Чикаго», «Ванкувер» тощо.
Також ви можете завантажити повний текст наукової публікації у форматі «.pdf» та прочитати онлайн анотацію до роботи, якщо відповідні параметри наявні в метаданих.
Переглядайте дисертації для різних дисциплін та оформлюйте правильно вашу бібліографію.
Verzelen, Nicolas. "Modèles graphiques gaussiens et sélection de modèles." Phd thesis, Université Paris Sud - Paris XI, 2008. http://tel.archives-ouvertes.fr/tel-00352802.
Повний текст джерелаEn utilisant le lien entre modèles graphiques et régression linéaire à plan d'expérience gaussien, nous développons une approche basée sur des techniques de sélection de modèles. Les procédures ainsi introduites sont analysés d'un point de vue non-asymptotique. Nous prouvons notamment des inégalités oracles et des propriétés d'adaptation au sens minimax valables en grande dimension. Les performances pratiques des méthodes statistiques sont illustrées sur des données simulées ainsi que sur des données réelles.
Gaudel, Romaric. "Paramètres d'ordre et sélection de modèles en apprentissage : caractérisation des modèles et sélection d'attributs." Phd thesis, Université Paris Sud - Paris XI, 2010. http://tel.archives-ouvertes.fr/tel-00549090.
Повний текст джерелаArlot, Sylvain. "Rééchantillonnage et Sélection de modèles." Phd thesis, Université Paris Sud - Paris XI, 2007. http://tel.archives-ouvertes.fr/tel-00198803.
Повний текст джерелаLa majeure partie de ce travail de thèse consiste dans la calibration précise de méthodes de sélection de modèles optimales en pratique, pour le problème de la prédiction. Nous étudions la validation croisée V-fold (très couramment utilisée, mais mal comprise en théorie, notamment pour ce qui est de choisir V) et plusieurs méthodes de pénalisation. Nous proposons des méthodes de calibration précise de pénalités, aussi bien pour ce qui est de leur forme générale que des constantes multiplicatives. L'utilisation du rééchantillonnage permet de résoudre des problèmes difficiles, notamment celui de la régression avec un niveau de bruit variable. Nous validons théoriquement ces méthodes du point de vue non-asymptotique, en prouvant des inégalités oracle et des propriétés d'adaptation. Ces résultats reposent entre autres sur des inégalités de concentration.
Un second problème que nous abordons est celui des régions de confiance et des tests multiples, lorsque l'on dispose d'observations de grande dimension, présentant des corrélations générales et inconnues. L'utilisation de méthodes de rééchantillonnage permet de s'affranchir du fléau de la dimension, et d'"apprendre" ces corrélations. Nous proposons principalement deux méthodes, et prouvons pour chacune un contrôle non-asymptotique de leur niveau.
Liquet, benoit. "Sélection de modèles semi-paramétriques." Phd thesis, Université Victor Segalen - Bordeaux II, 2002. http://tel.archives-ouvertes.fr/tel-00002430.
Повний текст джерелаLiquet, Benoit. "Sélection de modèles semi-paramétriques." Bordeaux 2, 2002. http://www.theses.fr/2002BOR20958.
Повний текст джерелаCarrier, Vincent. "Modèles de sélection sexuelle à deux espèces." Thesis, National Library of Canada = Bibliothèque nationale du Canada, 1996. http://www.collectionscanada.ca/obj/s4/f2/dsk3/ftp04/nq21439.pdf.
Повний текст джерелаDurand, Jean-Baptiste. "Modèles à structure cachée : inférence, estimation, sélection de modèles et applications." Phd thesis, Université Joseph Fourier (Grenoble), 2003. https://tel.archives-ouvertes.fr/tel-00002754v3.
Повний текст джерелаKouyoumdjian, Alexandre. "Sélection de cibles en mouvement : contexte, modèles, et paradigmes d'aide à la sélection." Thesis, Université Paris-Saclay (ComUE), 2018. http://www.theses.fr/2018SACLS606/document.
Повний текст джерелаThe selection of moving targets has received little attention in the literature, as the factors that influence motion are numerous and complex. Though models such as Fitts’ Law can estimate selection time for static targets, they fail to do so for moving ones, and the influence of a target’s movements on its associated selection performance remains to be determined. Here we propose a state of the art of moving target selection techniques, a taxonomy thereof, as well as a classification of moving targets and their environments. We propose a model for the description and generation of movement, a use it to extract essential motion parameters from moving targets, in order to estimate their selection difficulty. This model (SFA) includes speed (S), the period between each change in direction and its associated frequency (F), as well as the maximum angular amplitude of these changes (A).Thanks to the SFA model, we measured the influence of these parameters on selection performance, the subjective impressions of users, and their anticipation strategies. These results led us to look for SFA dependent criteria, such as the area of the target’s trajectory’s convex hull over a given period of time, which can predict selection performance. We note that Fitts’s distance has little influence on selection performance for quick, unpredictable moving targets, and validate our estimate of selection performance by showing that this estimate can be used to improve selection performance by adjusting the size of targets accordingly, or by using it to bias an intention prediction heuristic. We also assess the benefits of pseudo-haptic assistance coupled with intention prediction, and show that it depends on the speed accuracy trade-off chosen by a given user. We finally show that a technique based on the manual elimination of distractors running concurrently with intention prediction allows for excellent selection performance, and drastically reduced physical effort. We conclude by offering advice on the design of new selection techniques that would be better suited to moving targets
Donnet, Sophie. "Inversion de données IRMf : estimation et sélection de modèles." Paris 11, 2006. http://www.theses.fr/2006PA112193.
Повний текст джерелаThis thesis is devoted to the analysis of functional Magnetic Resonance Imaging data (fMRI). In the framework of standard convolution models, we test a model that allows for the variation of the magnitudes of the hemodynamic reponse. To estimate the parameters of this model, we have to resort to the Expectation-Maximisation algorithm. We test this model against the standard one --withconstant magnitudes-- on several real data, set by a likelihood ratio test. The linear model suffers from a lack of biological basis, hence we consider a physiological model. In this framework, we describe the data as the sum of a regression term, defined as the non-analytical solution of an ordinary differentiel equation (ODE) depending on random parameters, and a gaussian observation noise. We develop a general method to estimate the parameters of a statistical model defined by ODE with non-observed parameters. This method, integrating a numerical resolution of the ODE, relies on a stochastic version of the EM algorithm. The convergence of the algorithm is proved and the error induced by the numerical solving method is controlled. We apply this method on simulated and real data sets. Subsequently, we consider statistical models defined by stochastic differential equations (SDE) depending on random parameters. We approximate the diffusion process by a numerical scheme and propose a general estimation method. Results of a pharmacokineticmixed model study (on simulated and real data set) illustrate the accuracy of the estimation and the relevance of the SDE approach. Finally, the identifiability of statistical models defined by SDE with random parameters is studied
Vandewalle, Vincent. "Estimation et sélection en classification semi-supervisée." Phd thesis, Université des Sciences et Technologie de Lille - Lille I, 2009. http://tel.archives-ouvertes.fr/tel-00447141.
Повний текст джерелаBertrand, Julie. "Pharmacogénétique en Pharmacocinétique de population : tests et sélection de modèles." Phd thesis, Université Paris-Diderot - Paris VII, 2009. http://tel.archives-ouvertes.fr/tel-00482994.
Повний текст джерелаMattei, Pierre-Alexandre. "Sélection de modèles parcimonieux pour l’apprentissage statistique en grande dimension." Thesis, Sorbonne Paris Cité, 2017. http://www.theses.fr/2017USPCB051/document.
Повний текст джерелаThe numerical surge that characterizes the modern scientific era led to the rise of new kinds of data united in one common immoderation: the simultaneous acquisition of a large number of measurable quantities. Whether coming from DNA microarrays, mass spectrometers, or nuclear magnetic resonance, these data, usually called high-dimensional, are now ubiquitous in scientific and technological worlds. Processing these data calls for an important renewal of the traditional statistical toolset, unfit for such frameworks that involve a large number of variables. Indeed, when the number of variables exceeds the number of observations, most traditional statistics becomes inefficient. First, we give a brief overview of the statistical issues that arise with high-dimensional data. Several popular solutions are presented, and we present some arguments in favor of the method utilized and advocated in this thesis: Bayesian model uncertainty. This chosen framework is the subject of a detailed review that insists on several recent developments. After these surveys come three original contributions to high-dimensional model selection. A new algorithm for high-dimensional sparse regression called SpinyReg is presented. It compares favorably to state-of-the-art methods on both real and synthetic data sets. A new data set for high-dimensional regression is also described: it involves predicting the number of visitors in the Orsay museum in Paris using bike-sharing data. We focus next on model selection for high-dimensional principal component analysis (PCA). Using a new theoretical result, we derive the first closed-form expression of the marginal likelihood of a PCA model. This allows us to propose two algorithms for model selection in PCA. A first one called globally sparse probabilistic PCA (GSPPCA) that allows to perform scalable variable selection, and a second one called normal-gamma probabilistic PCA (NGPPCA) that estimates the intrinsic dimensionality of a high-dimensional data set. Both methods are competitive with other popular approaches. In particular, using unlabeled DNA microarray data, GSPPCA is able to select genes that are more biologically relevant than several popular approaches
Lerasle, Matthieu. "Rééchantillonnage et sélection de modèles optimale pour l'estimation de la densité." Toulouse, INSA, 2009. http://eprint.insa-toulouse.fr/archive/00000290/.
Повний текст джерелаPlancade, Sandra. "Estimation par sélection de modèles à partir de données partiellement observées." Paris 5, 2010. http://www.theses.fr/2010PA05S008.
Повний текст джерелаThis manuscript presents several non parametric estimation procedures in frameworks involving partially observed data. The estimators rely on the model selection method adapted to the L2 risk (following Birge and Massart procedure) and also to the risk at a given point. The first part of the manuscript is devoted to the estimation of regression error density, and the second part to survival analysis issues: estimation of the hazard rate in presence of right censoring and estimation of the conditional distribution function from interval censored data
Baraud, Yannick. "Sélection de modèles et estimation adaptative dans différents cadres de régression." Paris 11, 1998. http://www.theses.fr/1998PA112002.
Повний текст джерелаCaron, François. "Inférence bayésienne pour la détermination et la sélection de modèles stochastiques." Ecole Centrale de Lille, 2006. http://www.theses.fr/2006ECLI0012.
Повний текст джерелаWe are interested in the addition of uncertainty in hidden Markov models. The inference is made in a Bayesian framework based on Monte Carlo methods. We consider multiple sensors that may switch between several states of work. An original jump model is developed for different kind of situations, including synchronous/asynchronous data and the binary valid/invalid case. The model/algorithm is applied to the positioning of a land vehicle equipped with three sensors. One of them is a GPS receiver, whose data are potentially corrupted due to multipaths phenomena. We consider the estimation of the probability density function of the evolution and observation noises in hidden Markov models. First, the case of linear models is addressed and MCMC and particle filter algorithms are developed and applied on three different applications. Then the case of the estimation of probability density functions in nonlinear models is addressed. For that purpose, time-varying Dirichlet processes are defined for the online estimation of time-varying probability density functions
André, Jean-Baptiste. "Niveaux de sélection chez les microparasites : virulence, coopération, mutation." Montpellier 2, 2003. http://www.theses.fr/2003MON20172.
Повний текст джерелаSauvé, Marie. "Sélection de modèles en régression non gaussienne : applications à la sélection de variables et aux tests de survie accélérés." Paris 11, 2006. http://www.theses.fr/2006PA112201.
Повний текст джерелаThis thesis deals with model selection in non Gaussian regression. Our aim is to get informations on a function s given only some values perturbed by noises non necessarily Gaussian. In a first part, we consider histogram models (i. E. Classes of piecewise constant functions) associated with a collection of partitions of the set on which s is defined. We determine a penalized least squares criterion which selects a partition whose associated estimator satisfies an oracle inequality. Selecting a histogram model does not always lead to an accurate estimation of s, but allows for example to detect the change-points of s. In order to perform variable selection, we also propose a non linear method which relies on the use of CART and on histogram model selection. In a second part, we consider piecewise polynomial models, whose approximation properties are better. We aim at estimating s with a piecewise polynomial whose degree can vary from region to region. We determine a penalized criterion which selects a partition and a series of degrees whose associated piecewise polynomial estimator satisfies an oracle inequality. We also apply this result to detect the change-points of a piecewise affine function. The aim of this last work is to provide an adequate stress interval for Accelerating Life Test
Boisbunon, Aurélie. "Sélection de modèle : une approche décisionnelle." Phd thesis, Université de Rouen, 2013. http://tel.archives-ouvertes.fr/tel-00793898.
Повний текст джерелаDehouche, Nassim. "Management de portefeuilles de projets : modèles multicritère d'évaluation, de sélection et d'argumentation." Thesis, Paris 9, 2014. http://www.theses.fr/2014PA090017.
Повний текст джерелаProject portfolio management (PPM) involves the use of methods and tools, allowing an organization to plan, evaluate, analyze and screen the execution of a set of projects or project proposals, sharing common resources or aiming at the attainment of common objectives. Multicriteria decision aid models are useful tools to support this process, given their ability to accurately model preferences, and rationally agregate points of view. However, existing models present some lacks that limit their use outside of academic circles : (i) They neglect the non-symetrical nature of the importance of some criteria that are relevant in PPM. (ii) The black box effect makes it hard to use them for the argumentation of decisions and to gain their acceptance by users (iii) They are implicitly fitted for private/for-profit projects, which limits their use in public organizations. In this thesis, our contribution consists in proposing two multicriteria methods for supporting the activities of evaluating, selecting and arguing decision, for project portfolio management. We propose: (i) An analysis of the specific features of public and private projects and their consequences for decision support (ii) A framework that allows an independent modeling of the abilities of a criterion to oppose and convince (iii) Two transparent multicriteria agregation procedures, fitted for different decision contexts. We ensure the theoretical validity of our approaches and illustrate their applicability on real data, with satisfying results
Harroue, Benjamin. "Approche bayésienne pour la sélection de modèles : application à la restauration d’image." Thesis, Bordeaux, 2020. http://www.theses.fr/2020BORD0127.
Повний текст джерелаInversing main goal is about reconstructing objects from data. Here, we focus on the special case of image restauration in convolution problems. The data are acquired through a altering observation system and additionnaly distorted by errors. The problem becomes ill-posed due to the loss of information. One way to tackle it is to exploit Bayesian approach in order to regularize the problem. Introducing prior information about the unknown quantities osset the loss, and it relies on stochastic models. We have to test all the candidate models, in order to select the best one. But some questions remain : how do you choose the best model? Which features or quantities should we rely on ? In this work, we propose a method to automatically compare and choose the model, based on Bayesion decision theory : objectively compare the models based on their posterior probabilities. These probabilities directly depend on the marginal likelihood or “evidence” of the models. The evidence comes from the marginalization of the jointe law according to the unknow image and the unknow hyperparameters. This a difficult integral calculation because of the complex dependancies between the quantities and the high dimension of the image. That way, we have to work with computationnal methods and approximations. There are several methods on the test stand as Harmonic Mean, Laplace method, discrete integration, Chib from Gibbs approximation or the power posteriors. Comparing is those methods is significative step to determine which ones are the most competent in image restauration. As a first lead of research, we focus on the family of Gaussian models with circulant covariance matrices to lower some difficulties
Villers, Fanny. "Tests et sélection de modèles pour l'analyse de données protéomiques et transcriptomiques." Paris 11, 2007. http://www.theses.fr/2007PA112198.
Повний текст джерелаThe techniques for gathering data expression for a large number of genes or proteins have grown in recent years. The purpose of this thesis is to provide statistical methods appropriate to treat these data. The first part concerns the differential analysis of proteomic data obtained from bidimensional electrophoresis. The aim is to detect the proteins whose abundance differs according to the experimental condition. When we compare simultaneously more than two conditions, this comes to detect the non-zero components of the mean of a Gaussian vector whose components are not independent and whose dependence structure is known. We propose a model selection approach based on the minimization of a penalized least squares criterion. The two other parts of the thesis concern Gaussian graphical models, that can be used to desribe interactions between genes. In the second part we propose a study based on simulations to compare the performances of several methods of graph estimation. In the third part we propose a test of graph. Indeed, biologists often have a previous knowledge of the genetic network and may want to assess the quality of their model thanks to gene expression data. To this aim we constructed a procedure for testing the neighborhoods of the nodes of the graph. Our procedure is based on the test of a linear hypothesis in a Gaussian linear regression in random Gaussian design
Khadraoui, Lobna. "Sélection de copules archimédiennes dans un modèle semi-paramétrique." Master's thesis, Université Laval, 2018. http://hdl.handle.net/20.500.11794/30251.
Повний текст джерелаThis work considers a semi-parametric linear model with error terms modeled by a copula chosen from the Archimedean family or the normal copula. The modeling of errors by a copula provides flexibility and makes it possible to characterize the dependency structure in a simple and effective manner. The simplicity lies in the fact that a single parameter α controls the degree of dependency present in the data. The efficiency is in the fact that this semi-parametric model weakens standard assumptions often encountered in applied statistics namely normality and independence. After an implementation of the model based on a copula we proposed a theoretical study on the asymptotic behavior of the estimator of the dependence parameter α by showing its consistency and its asymptotic normality under classical assumptions of regularity. Estimation of the model parameters is performed by maximizing a pseudo-likelihood. The selection of the best copula that fits the data for each case is based on the Akaike selection criterion. A comparison with the criterion of cross-validation is presented as well. Finally, a numerical study on simulated and real data sets is proposed.
Bah, Boubacar. "Le modèle du Look-down avec sélection." Thesis, Aix-Marseille, 2012. http://www.theses.fr/2012AIXM4711/document.
Повний текст джерелаThe purpose of the dissertation is to study the look-down model with selection in the case of a population composed only two alleles, one of them has a selective advantage. In this thesis, this selective advantage is modelled by a death rate for the wild-type allele. In the first part, we are interested in the case of a population of infinite size. We show the model is well defined. We show convergence in probability, as the population size tends to infinity, towards the Wright-Fisher diffusion with selection. In the second part we study a variant of the simplest look-down with selection where the size of the population is finite and fixed. We propose two methods of convergence of this finite model towards the Wright-Fisher diffusion with selection. Finally, another approach is considered. We study the look-down model with selection when we replace the usual reproduction model, which is dual to Kingman's coalescent by a population model dual to the Lambda-coalescent in the case of a population of infinite size. We first show this model is well defined. We show that the proportion of one of the two types converges in probability, as the population size N tends to infinity, towards the solution of a stochastic differential equation driven by a Poisson point process. Finally, we show that one of the two types fixate in finite time if and only if the Lambda-coalescent comes down from infinity
Feki, Yassine. "Opérationnalisation des stratégies de sélection des prestataires logistiques." Thesis, Université Laval, 2013. http://www.theses.ulaval.ca/2013/29826/29826.pdf.
Повний текст джерелаSmadi, Charline. "Modèles probabilistes de populations : branchement avec catastrophes et signature génétique de la sélection." Thesis, Paris Est, 2015. http://www.theses.fr/2015PESC1035/document.
Повний текст джерелаThis thesis is devoted to the probabilistic study of demographic and genetical responses of a population to some point wise events. In a first part, we are interested in the effect of random catastrophes, which kill a fraction of the population and occur repeatedly, in populations modeled by branching processes. First we construct a new class of processes, the continuous state branching processes with catastrophes, as the unique strong solution of a stochastic differential equation. Then we describe the conditions for the population extinction. Finally, in the case of almost sure absorption, we state the asymptotical rate of absorption. This last result has a direct application to the determination of the number of infected cells in a model of cell infection by parasites. Indeed, the parasite population size in a lineage follows in this model a branching process, and catastrophes correspond to the sharing of the parasites between the two daughter cells when a division occurs. In a second part, we focus on the genetic signature of selective sweeps. The genetic material of an individual (mostly) determines its phenotype and in particular some quantitative traits, as birth and intrinsic death rates, and interactions with others individuals. But genotype is not sufficient to determine "adaptation" in a given environment: for example the life expectancy of a human being is very dependent on his environment (access to drinking water, to medical infrastructures,...). The eco-evolutive approach aims at taking into account the environment by modeling interactions between individuals. When a mutation or an environmental modification occurs, some alleles can invade the population to the detriment of other alleles: this phenomenon is called a selective sweep and leaves signatures in the neutral diversity in the vicinity of the locus where the allele fixates. Indeed, this latter "hitchhiking” alleles situated on loci linked to the selected locus. The only possibility for an allele to escape this "hitchhiking" is the occurrence of a genetical recombination, which associates it to another haplotype in the population. We quantify the signature left by such a selective sweep on the neutral diversity. We first focus on neutral proportion variation in loci partially linked with the selected locus, under different scenari of selective sweeps. We prove that these different scenari leave distinct signatures on neutral diversity, which can allow to discriminate them. Then we focus on the linked genealogies of two neutral alleles situated in the vicinity of the selected locus. In particular, we quantify some statistics under different scenari of selective sweeps, which are currently used to detect recent selective events in current population genetic data. In these works the population evolves as a multitype birth and death process with competition. If such a model is more realistic than branching processes, the non-linearity caused by competitions makes its study more complex
Baudry, Jean-Patrick. "Sélection de modèle pour la classification non superviséeChoix du nombre de classes." Paris 11, 2009. http://www.theses.fr/2009PA112265.
Повний текст джерелаThe reported works take place in the statistical framework of model-based clustering. We particularly focus on choosing the number of classes and on the ICL model selection criterion. A fruitful approach for theoretically studying it consists of considering a contrast related to the clustering purpose. This entails the definition and study of a new estimator and new model selection criteria. Practical solutions are provided to compute them, which can also be applied to the computation of the usual maximum likelihood estimator within mixture models. The slope heuristics is applied to the calibration of the considered penalized criteria. Thus its theoretical bases are recalled in details and two approaches for its application are studied. Another approach for model-based clustering is considered: each class itself may be modeled by a Gaussian mixture. A methodology is proposed, notably to tackle the question of which components have to be merged. Finally a criterion is proposed, which enables to choose a number of components ---~when identified to the number of classes~--- related to a known external classification
Gendre, Xavier. "Estimation par sélection de modèle en régression hétéroscédastique." Phd thesis, Université de Nice Sophia-Antipolis, 2009. http://tel.archives-ouvertes.fr/tel-00397608.
Повний текст джерелаLa première partie de cette thèse consiste dans l'étude du problème d'estimation de la moyenne et de la variance d'un vecteur gaussien à coordonnées indépendantes. Nous proposons une méthode de choix de modèle basée sur un critère de vraisemblance pénalisé. Nous validons théoriquement cette approche du point de vue non-asymptotique en prouvant des majorations de type oracle du risque de Kullback de nos estimateurs et des vitesses de convergence uniforme sur les boules de Hölder.
Un second problème que nous abordons est l'estimation de la fonction de régression dans un cadre hétéroscédastique à dépendances connues. Nous développons des procédures de sélection de modèle tant sous des hypothèses gaussiennes que sous des conditions de moment. Des inégalités oracles non-asymptotiques sont données pour nos estimateurs ainsi que des propriétés d'adaptativité. Nous appliquons en particulier ces résultats à l'estimation d'une composante dans un modèle de régression additif.
Fauchart, Emmanuelle. "Deux formalisations des processus de sélection en théorie économique." Paris 1, 1996. http://www.theses.fr/1996PA010062.
Повний текст джерелаThe thesis identifies two different mechanisms of selection : selection can perform either in the form of successive shocks perturbating a space of proportions or in the form of some simultaneous evaluation of the competing entities. Whereas the first mechanism is indissociable from increasing returns in order to make the collective structure converge towards some stable proportions, the second mechanism can be associated either with increasing or ( and) decreasing returns type of properties. We propose that the action of selection through either one of these two mechanisms is indissociable from the emergence of intertemporal effects giving rise to irreversibilities. These irreversibilities are the basis of the orientation and organization function of selection in evolutionary dynamics. These intertemporal effects can either be individual or( and) collective. Collective effects bear more irreversibilities in the dynamics because they lie on interdependencies among the members of the population (adopters or firms) impulsing the evolution. Besides intertemporal interdependencies arising from the sequencing of time, some other factors orienting evolutionary dynamics are identified
Baey, Charlotte. "Modélisation de la variabilité inter-individuelle dans les modèles de croissance de plantes et sélection de modèles pour la prévision." Phd thesis, Ecole Centrale Paris, 2014. http://tel.archives-ouvertes.fr/tel-00985747.
Повний текст джерелаTran, Phuoc Nguyen. "Modèles de sélection d'interface et d'association flux/interface pour les terminaux mobiles multi-interfaces." Phd thesis, Télécom ParisTech, 2010. http://pastel.archives-ouvertes.fr/pastel-00564095.
Повний текст джерелаSokolovska, Nataliya. "Contributions à l'estimation de modèles probabilistes discriminants : apprentissage semi-supervisé et sélection de caractéristiques." Phd thesis, Ecole nationale supérieure des telecommunications - ENST, 2010. http://tel.archives-ouvertes.fr/tel-00557662.
Повний текст джерелаSaumard, Adrien. "Estimation par Minimum de Contraste Régulier et Heuristique de Pente en Sélection de Modèles." Phd thesis, Université Rennes 1, 2010. http://tel.archives-ouvertes.fr/tel-00569372.
Повний текст джерелаOllier, Edouard. "Sélection de modèles statistiques par méthodes de vraisemblance pénalisée pour l'étude de données complexes." Thesis, Lyon, 2017. http://www.theses.fr/2017LYSEN097.
Повний текст джерелаThis thesis is mainly devoted to the development of penalized maximum likelihood methods for the study of complex data.A first work deals with the selection of generalized linear models in the framework of stratified data, characterized by the measurement of observations as well as covariates within different groups (or strata). The purpose of the analysis is then to determine which covariates influence in a global way (whatever the stratum) the observations but also to evaluate the heterogeneity of this effect across the strata.Secondly, we are interested in the selection of nonlinear mixed effects models used in the analysis of longitudinal data. In a first work, we describe a SAEM-type algorithm in which the penalty is taken into account during step M by solving a penalized regression problem at each iteration. In a second work, inspired by proximal gradient algorithms, we simplify the M step of the penalized SAEM algorithm previously described by performing only one proximal gradient iteration at each iteration. This algorithm, called Stochastic Approximation Proximal Gradient Algorithm (SAPG), corresponds to a proximal gradient algorithm in which the gradient of the likelihood is approximated by a stochastic approximation technique.Finally, we present two statistical modeling works realized during this thesis
Truntzer, Caroline. "Évaluation de modèles pronostiques issus de l'analyse dutranscriptome." Phd thesis, Université Claude Bernard - Lyon I, 2007. http://tel.archives-ouvertes.fr/tel-00161161.
Повний текст джерелаétude soulève cependant de nombreuses questions statistiques dues à l'analyse simultanée de l'expression
de milliers de gènes pour un nombre restreint de patients. Nous nous sommes intéressés
à deux aspects de l'évaluation des modèles pronostiques issus de l'analyse du transcriptome. Dans
un premier temps, l'utilisation complémentaire de jeux de données simulés et publics nous a permis
de montrer que le choix de la méthode d'analyse la plus adaptée repose sur la manière dont ses
propriétés théoriques s'adaptent à la structure des données. Cette réflexion a été appliquée aux
analyses discriminante et inter-groupes. Dans un second temps, des simulations nous ont permis
d'estimer les contributions respectives des variables clinico-biologiques classiques et transcriptomiques
dans des modèles de survie. Les paramètres associés à la surestimation de la contribution
des biopuces ont été identifiés.
Sidi, Zakari Ibrahim. "Sélection de variables et régression sur les quantiles." Thesis, Lille 1, 2013. http://www.theses.fr/2013LIL10081/document.
Повний текст джерелаThis work is a contribution to the selection of statistical models and more specifically in the selection of variables in penalized linear quantile regression when the dimension is high. It focuses on two points in the selection process: the stability of selection and the inclusion of variables by grouping effect. As a first contribution, we propose a transition from the penalized least squares regression to quantiles regression (QR). A bootstrap approach based on frequency of selection of each variable is proposed for the construction of linear models (LM). In most cases, the QR approach provides more significant coefficients. A second contribution is to adapt some algorithms of "Random" LASSO (Least Absolute Shrinkage and Solution Operator) family in connection with the QR and to propose methods of selection stability. Examples from food security illustrate the obtained results. As part of the penalized QR in high dimension, the grouping effect property is established under weak conditions and the oracle ones. Two examples of real and simulated data illustrate the regularization paths of the proposed algorithms. The last contribution deals with variable selection for generalized linear models (GLM) using the nonconcave penalized likelihood. We propose an algorithm to maximize the penalized likelihood for a broad class of non-convex penalty functions. The convergence property of the algorithm and the oracle one of the estimator obtained after an iteration have been established. Simulations and an application to real data are also presented
Bouquet, Alban. "Amélioration de l'efficacité des programmes de sélection des bovins allaitants : de nouveaux modèles d'évaluation génétique." Phd thesis, AgroParisTech, 2009. http://pastel.archives-ouvertes.fr/pastel-00005713.
Повний текст джерелаBouquet, Alban. "Amélioration de l'efficacité des programmes de sélection des bovins allaitants : de nouveaux modèles d’évaluation génétique." Paris, AgroParisTech, 2009. http://pastel.paristech.org/5713/01/Thèse_ABouquet_14janv2010.pdf.
Повний текст джерелаTran, Phuoc Nguyen. "Modèles de sélection d'interface et d'association de flux/interface pour les terminaux mobiles multi-homés." Phd thesis, Paris, Télécom ParisTech, 2010. https://pastel.hal.science/pastel-00564095.
Повний текст джерелаThe diversity of radio access technologies (e. G. , GPRS, UMTS, HSDPA,Wi-Fi, WiMAX, LTE…), their complementary in terms of coverage area, technical characteristics (e. G. , bandwidth, QoS) and commercial opportunities for the operators lead to the development of mobile terminals integrating multiple radio interfaces. The ability of mobile terminals to support various interfaces provides many interesting benefits, such as permanent and ubiquitous access, reliability, load sharing/load balancing, bandwidth aggregation, and muti-criteria interface selection. Mobile terminals with several radio interfaces have the possibility to choose the ―best‖ interface according to several parameters such as application characteristics, user preferences, network characteristics, operator policies, tariff constraints, etc. It becomes also possible to associate the applications to the available network interfaces basing mainly on application requirements. In the thesis, we tackle the interface selection issue where a mobile terminal equipped with several interfaces has to select at any time the best interface or the best access technology according to multiple criteria. We particularly focus on the decision schemes and investigate the MADM methods. The fundamental objective of the MADM methods is to determine among a finite set of alternatives the optimal one. MADM includes many methods such as Simple Additive Weighting (SAW), Weighting Product (WP), and Technique for Order Preference by Similarity to Ideal Solution (TOPSIS)
Zulian, Marine. "Méthodes de sélection et de validation de modèles à effets mixtes pour la médecine génomique." Thesis, Institut polytechnique de Paris, 2020. http://www.theses.fr/2020IPPAX003.
Повний текст джерелаThe study of complex biological phenomena such as human pathophysiology, pharmacokinetics of a drug or its pharmacodynamics can be enriched by modelling and simulation approaches. Technological advances in genetics allow the establishment of data sets from larger and more heterogeneous populations. The challenge is then to develop tools that integrate genomic and phenotypic data to explain inter-individual variability. In this thesis, we develop methods that take into account the complexity of biological data and the complexity of underlying processes. Curation steps of genomic covariates allow us to limit the number of potential covariates and limit correlations between covariates. We propose an algorithm for selecting covariates in a mixed effects model whose structure is constrained by the physiological process. In particular, we illustrate the developed methods on two medical applications: actual high blood pressure data and simulated tramadol (opioid) metabolism data
Lavarde, Marc. "Fiabilité des semi-conducteurs, tests accélérés, sélection de modèles définis par morceaux et détection de sur-stress." Paris 11, 2007. http://www.theses.fr/2007PA112266.
Повний текст джерелаThis thesis deals with the using of accelerating data and regression model selection for high technology field: semiconductor chips. The accelerating trail gives us regression frameworks. The aim of the accelerating test consists on fitting the logarithm of the lifetime through the use of some function f, called the acceleration function. However, accelerating data may have misleading and complex comportment. In order to adapt the model with such data, we have proposed to detect the changes on the comportment of the acceleration function. We have considered a collection of piecewise acceleration models candidate to the estimation. For each model candidate we have estimated the least-squares estimation. And we have selected the final estimator using a penalized criterion. The penalized estimator is optimal approximation of the reality since the quadratic risk of penalized estimator is bounded by the minimal risk upon every least-squares estimators candidates. Moreover, this oracle inequality is non asymptotic. Furthermore, we have considered classical reliability cases: the Lognormal case associating with some fatigue failure, and the Weibull case associating with some choc failure. Lastly we have implemented model selection tools in order to realise survey study without a priori on the acceleration models and to use overstress trials
Avalos, Marta. "Modèles additifs parcimonieux." Phd thesis, Université de Technologie de Compiègne, 2004. http://tel.archives-ouvertes.fr/tel-00008802.
Повний текст джерелаMartinez, Marie-José. "Modèles linéaires généralisés à effets aléatoires : contributions au choix de modèle et au modèle de mélange." Phd thesis, Université Montpellier II - Sciences et Techniques du Languedoc, 2006. http://tel.archives-ouvertes.fr/tel-00388820.
Повний текст джерелаGuilloux, Agathe. "Inférence non paramétrique en statistique des durées de vie sous biais de sélection." Rennes 1, 2004. http://www.theses.fr/2004REN10058.
Повний текст джерелаRohart, Florian. "Prédiction phénotypique et sélection de variables en grande dimension dans les modèles linéaires et linéaires mixtes." Thesis, Toulouse, INSA, 2012. http://www.theses.fr/2012ISAT0027/document.
Повний текст джерелаRecent technologies have provided scientists with genomics and post-genomics high-dimensional data; there are always more variables that are measured than the number of individuals. These high dimensional datasets usually need additional assumptions in order to be analyzed, such as a sparsity condition which means that only a small subset of the variables are supposed to be relevant. In this high-dimensional context we worked on a real dataset which comes from the pig species and high-throughput biotechnologies. Metabolomic data has been measured with NMR spectroscopy and phenotypic data has been mainly obtained post-mortem. There are two objectives. On one hand, we aim at obtaining good prediction for the production phenotypes and on the other hand we want to pinpoint metabolomic data that explain the phenotype under study. Thanks to the Lasso method applied in a linear model, we show that metabolomic data has a real prediction power for some important phenotypes for livestock production, such as a lean meat percentage and the daily food consumption. The second objective is a problem of variable selection. Classic statistical tools such as the Lasso method or the FDR procedure are investigated and new powerful methods are developed. We propose a variable selection method based on multiple hypotheses testing. This procedure is designed to perform in linear models and non asymptotic results are given under a condition on the signal. Since supplemental data are available on the real dataset such as the batch or the family relationships between the animals, linear mixed models are considered. A new algorithm for fixed effects selection is developed, and this algorithm turned out to be faster than the usual ones. Thanks to its structure, it can be combined with any variable selection methods built for linear models. However, the convergence property of this algorithm depends on the method that is used. The multiple hypotheses testing procedure shows good empirical results. All the mentioned methods are applied to the real data and biological relationships are emphasized
Thouvenot, Vincent. "Estimation et sélection pour les modèles additifs et application à la prévision de la consommation électrique." Thesis, Université Paris-Saclay (ComUE), 2015. http://www.theses.fr/2015SACLS184/document.
Повний текст джерелаFrench electricity load forecasting encounters major changes since the past decade. These changes are, among others things, due to the opening of electricity market (and economical crisis), which asks development of new automatic time adaptive prediction methods. The advent of innovating technologies also needs the development of some automatic methods, because we have to study thousands or tens of thousands time series. We adopt for time prediction a semi-parametric approach based on additive models. We present an automatic procedure for covariate selection in a additive model. We combine Group LASSO, which is selection consistent, with P-Splines, which are estimation consistent. Our estimation and model selection results are valid without assuming that the norm of each of the true non-zero components is bounded away from zero and need only that the norms of non-zero components converge to zero at a certain rate. Real applications on local and agregate load forecasting are provided.Keywords: Additive Model, Group LASSO, Load Forecasting, Multi-stage estimator, P-Splines, Variables selection
Jin, Yinfu. "Identification Les paramètres des sols et sélection de modèles de comportement en utilisant des algorithmes génétiques." Thesis, Ecole centrale de Nantes, 2016. http://www.theses.fr/2016ECDN0017.
Повний текст джерелаThe subject of this thesis is the identification of soil parameters and the selection of constitutive models using genetic algorithms. First, various optimization methods for identifying soil parameters are studied. Then, a real - coded genetic algorithm (RCGA) has been developed to improve the performance of genetic algorithms (GA) for identifying soil parameters. Subsequently, the RCG A is employed to construct a formula for predicting the compressibility of remolded clays by using an evolutionary polynomial regression ( EPR ) based on the initial void ratio e 0 , the liquid limit w L and the plastic index I P . Then, an efficient procedure fo r identifying the necessary parameters of soft structured clay s is propose d by employing the enhanced RCGA coupled with an advanced anisotropic elasto - viscoplastic model. This approach is then validated and several applications are developed to demonstrate that the procedure can be used with a reduction of the testing cost . F inally , an appropriate model of sand with the necessary features based on conventional tests and with an easy way of identifying parameters for geotechnical applications by employ ing th e RCGA and different sand models is selected. A discussion on nonlinear plastic stress - strain hardening , the incorporation of the critical state concept with interlocking effect , test types and numbers , and necessary strain level for the selection and use of sand models concludes the thesis
Bourguignon, Pierre Yves Vincent. "Parcimonie dans les modèles Markoviens et application à l'analyse des séquences biologiques." Thesis, Evry-Val d'Essonne, 2008. http://www.theses.fr/2008EVRY0042.
Повний текст джерелаMarkov chains, as a universal model accounting for finite memory, discrete valued processes, are omnipresent in applied statistics. Their applications range from text compression to the analysis of biological sequences. Their practical use with finite samples, however, systematically require to draw a compromise between the memory length of the model used, which conditions the complexity of the interactions the model may capture, and the amount of information carried by the data, whose limitation negatively impacts the quality of estimation. Context trees, as an extension of the model class of Markov chains, provide the modeller with a finer granularity in this model selection process, by allowing the memory length to vary across contexts. Several popular modelling methods are based on this class of models, in fields such as text indexation of text compression (Context Tree Maximization and Context Tree Weighting). We propose an extension of the models class of context trees, the Parcimonious context trees, which further allow the fusion of sibling nodes in the context tree. They provide the modeller with a yet finer granularity to perform the model selection task, at the cost of an increased computational cost for performing it. Thanks to a bayesian approach of this problem borrowed from compression techniques, we succeeded at desiging an algorithm that exactly optimizes the bayesian criterion, while it benefits from a dynamic programming scheme ensuring the minimisation of the computational complexity of the model selection task. This algorithm is able to perform in reasonable space and time on alphabets up to size 10, and has been applied on diverse datasets to establish the good performances achieved by this approach
Mellah, Kohi Meryem. "Modèle de caractérisation d'une bibliothèque CMOS : définition d'une sélection optimale d'éléments." Montpellier 2, 1995. http://www.theses.fr/1995MON20156.
Повний текст джерелаBekara, Maïza. "Optimisation de critères de choix de modèles pour un faible nombre de données." Paris 11, 2004. http://www.theses.fr/2004PA112139.
Повний текст джерелаIn this work we propose a model selection criterion based on Kullback's symmetric divergence. The developed criterion, called KICc is a bias corrected version of the asymptotic criterion KIC (Cavanaugh, Statistics and Probability Letters, vol. 42, 1999). The correction is of particular use when the sample size is small or when the number of fitted parameters is moderate to large fraction of the sample size. KICc is an exactly unbiased estimator for linear regression models and appreciatively unbiased for autoregressive and nonlinear regression models. The two criteria KIC and KICc are developed under the assumption that the true model is correctly specified or overfitted by the candidate models. We investigate the bias properties and the model selection performance of the two criteria in the underfitted case. An extension of KICc, called PKIC is also developed for the case of future experiment where date of interest is missing or indirectly observed. The KICc is implemented to solve the problem of denoising by using orthogonal projection and thresholding. The threshold is obtained as the absolute value of the kth largest coefficient that minimizes KICc. Finally, we propose a computational optimization of a cross validation based model selection criterion that uses the Bayesian predictive density as candidate model and marginal likelihood as a cost function. The developed criterion, CVBPD, is a consistent model selection criterion for linear regression