Teses / dissertações sobre o tema "Détection de modèles"
Crie uma referência precisa em APA, MLA, Chicago, Harvard, e outros estilos
Veja os 50 melhores trabalhos (teses / dissertações) para estudos sobre o assunto "Détection de modèles".
Ao lado de cada fonte na lista de referências, há um botão "Adicionar à bibliografia". Clique e geraremos automaticamente a citação bibliográfica do trabalho escolhido no estilo de citação de que você precisa: APA, MLA, Harvard, Chicago, Vancouver, etc.
Você também pode baixar o texto completo da publicação científica em formato .pdf e ler o resumo do trabalho online se estiver presente nos metadados.
Veja as teses / dissertações das mais diversas áreas científicas e compile uma bibliografia correta.
Ghorbanzadeh, Dariush. "Détection de rupture dans les modèles statistiques". Paris 7, 1992. http://www.theses.fr/1992PA077246.
Texto completo da fonteMoussi, Aurélie. "Détection de débris orbitaux : comparaison avec les modèles". Toulouse, ENSAE, 2005. http://www.theses.fr/2005ESAE0016.
Texto completo da fonteDang, Quoc Viet. "Similarités dans des Modèles BRep Paramétriques : Détection et Applications". Phd thesis, Toulouse, INPT, 2014. http://oatao.univ-toulouse.fr/12154/1/Dang_quoc_viet.pdf.
Texto completo da fonteLe, Gallou Sylvain. "Détection robuste des éléments faciaux par modèles actifs d'apparence". Rennes 1, 2007. http://www.theses.fr/2007REN1S083.
Texto completo da fonteFor man-machine interfaces, interactions with the machines in an unconstrained environment are a major issue. We use the Active Appearance Models (AAM) to precisely localize the eyes, the nose and the mouth of faces. Our work consists of making the AAM more robust to illumination, pose, identity and expression of faces. On the one hand, we propose a pretreatment based on oriented maps to get independent from the effects of illumination variations and on the other hand, an adaptive system allowing the AAM to focus itself in real time on the pre-learned specific model of a face the more adapted to the analyzed unknown face
Tabia, Karim. "Modèles graphiques et approches comportementales pour la détection d'intrusions". Artois, 2008. http://www.theses.fr/2008ARTO0407.
Texto completo da fonteIn this thesis, we deal with modelling intrusion detection problem using graphical models. We first define relevant variables taking into account nowadays attacks and normal traffic characteristics. We after that study the failure of standard Bayesian networks and decision trees in detecting novel attacks. We proposed two directions in order to solve this problem: the first one proposes to enhance and adapt standard Bayesian networks and decision trees in order to fit anomaly approach requirements and better detect novel attacks. The second direction proposes a serial combination aiming at equipping a classifier with an anomaly approach and a diagnosis component. We finally deal with analyzing audit events when inputs are uncertain or missing. We start with analyzing Jeffrey's rule for revising possibility distributions by uncertain observations. Then, we propose an efficient algorithm for revising possibility distributions encoded by a naive possibilistic network. This algorithm is particularly suitable for classification with uncertain inputs since it allows classification in polynomial time using different efficient transformations of the initial naive possibilistic networks
Smith, Isabelle. "Détection d'une source faible : modèles et méthodes statistiques : application à la détection d'exoplanètes par imagerie directe". Phd thesis, Université de Nice Sophia-Antipolis, 2010. http://tel.archives-ouvertes.fr/tel-00548905.
Texto completo da fonteSoule, Augustin. "Méthodes et modèles de détection d'anomalies dans les réseaux d'opérateurs". Paris 6, 2006. http://www.theses.fr/2006PA066088.
Texto completo da fonteMeacher, Duncan. "Binaires compactes : modèles de populations, détection multi-messagers et cosmologie". Thesis, Nice, 2015. http://www.theses.fr/2015NICE4058/document.
Texto completo da fonteHere I present my thesis investigating gravitational-wave data analysis in the following two areas. The first is to test the readiness of the LIGO-Virgo collaborations to the advanced detector era, which will begin in the summer of 2015, to make a detection of an astrophysical stochastic gravitational-wave background. The second is to continue an investigation into the science potential of a conceived, third generation gravitational-wave detector, the Einstein Telescope, in terms of astrophysics and cosmology. Both of these are conducted with the use of mock data and science challenges which consists of the production of expected gravitational-wave detector data, containing a large number of sources, that are simulated using realists distributions
Kenaza, Tayeb. "Modèles graphiques probabilistes pour la corrélation d'alertes en détection d'intrusions". Thesis, Artois, 2011. http://www.theses.fr/2011ARTO0401/document.
Texto completo da fonteIn this thesis, we focus on modeling the problem of alert correlation based on probabilistic graphical models. Existing approaches either require a large amount of expert knowledge or use simple similarity measures which are not enough to detect coordinated attacks. We first proposed a new modeling for the alert correlation problem, based on naive Bayesian classifiers, which can learn the coordination between elementary attacks that contribute to the achievement of an attack scenario. Our model requires only a slight contribution of expert knowledge. It takes advantage of available data and provides efficient algorithms for detecting and predicting attacks scenario. Then we show how our alert correlation approach can be improved by taking into account contextual information encoded in description logics, particularly in the context of a cooperative intrusion detection. Finally, we proposed several evaluation measures for a naive Bayesian multi-classifiers. This is very important for evaluating our alert correlation approach because it uses a set of naive Bayesian classifiers to monitor multiple intrusion objectives simultaneously
Kenaza, Tayeb. "Modèles graphiques probabilistes pour la corrélation d'alertes en détection d'intrusions". Electronic Thesis or Diss., Artois, 2011. http://www.theses.fr/2011ARTO0401.
Texto completo da fonteIn this thesis, we focus on modeling the problem of alert correlation based on probabilistic graphical models. Existing approaches either require a large amount of expert knowledge or use simple similarity measures which are not enough to detect coordinated attacks. We first proposed a new modeling for the alert correlation problem, based on naive Bayesian classifiers, which can learn the coordination between elementary attacks that contribute to the achievement of an attack scenario. Our model requires only a slight contribution of expert knowledge. It takes advantage of available data and provides efficient algorithms for detecting and predicting attacks scenario. Then we show how our alert correlation approach can be improved by taking into account contextual information encoded in description logics, particularly in the context of a cooperative intrusion detection. Finally, we proposed several evaluation measures for a naive Bayesian multi-classifiers. This is very important for evaluating our alert correlation approach because it uses a set of naive Bayesian classifiers to monitor multiple intrusion objectives simultaneously
Benabdallah, Khalid. "Identification et détection de modèles non stationnaires. Application aux signaux EEG". Rouen, 1992. http://www.theses.fr/1992ROUES035.
Texto completo da fonteMehmel, Cherif. "Modélisation et commande d'un interféromètre pour la détection d'ondes gravitationnelles". Chambéry, 1998. http://www.theses.fr/1998CHAMS021.
Texto completo da fonteElomary, Youssef. "Modèles déformables et multirésolution pour la détection de contours de traitement d'images". Phd thesis, Université Joseph Fourier (Grenoble), 1994. http://tel.archives-ouvertes.fr/tel-00010656.
Texto completo da fonteNotre propos dans cette thèse est d'étudier ces modèles dans un environnement multirésolution.
Commençant par une étude des contours actifs à haute résolution, nous démontrons un théorème d'existence pour les contours actifs fermés et les contours actifs à extrémités libres. Nous présentons ensuite un nouveau modèle appelé la bulle déformable, qui a l'avantage d'avoir une représentation discrète, d'être relativement robuste au bruit et à la texture et d'agir par faibles déformations.
Ensuite nous étudions quelques techniques de multirésolution, en présentant les avantages et les inconvénients de chacune. A travers une proposition que nous avons montrée, nous établissons le lien entre la multirésolution et la notion de minimisation
d'énergie.
Enfin, nous terminons par une proposition originale qui consiste à faire coopérer les contours actifs et la multirésolution. Cette coopération s'aggrémente de plusieurs approches pour faire passer le contour du haut de la pyramide vers sa base. Elle
associe entre autres une factorisation du modèle des contours actifs, d'une part selon une démarche de type membrane effectuée à basse résolution, et d'autre part selon
une démarche de type plaque mince au travers des différentes résolutions supérieures permettant de réajuster le contour détecté jusqu'à la résolution initiale.
Vicente, David. "Modèles de Mumford-Shah pour la détection de structures fines en image". Thesis, Orléans, 2015. http://www.theses.fr/2015ORLE2055/document.
Texto completo da fonteThis thesis is a contribution to the fine tubular structures detection problem in a 2-D or 3-D image. We arespecifically interested in the case of angiographic images. The vessels are surrounded by noise and thenthe question is to segment precisely the blood network. The theoretical framework of our work is thecalculus of variations and we focus on the Mumford-Shah energy. Initially, this model is adapted to thedetection of volumetric structures extended in all directions of the image. The aim of this study is to buildan energy that favors sets which are extended in one direction, which is the case of fine tubes. Then, weintroduce a new unknown, a Riemannian metric, which captures the geometric structure at each point ofthe image and we give a new formulation of the Mumford-Shah energy adapted to this metric. Thecomplete analysis of this model is done: we prove that the associated problem of minimization is wellposed and we introduce an approximation by gamma-convergence more suitable for numerics. Eventually,we propose numerical experimentations adapted to this approximation
Colot, Olivier. "Apprentissage et détection automatique de changements de modèles : application aux signaux électroencéphalographiques". Rouen, 1993. http://www.theses.fr/1993ROUES012.
Texto completo da fontePoncet, Bénédicte N. "Modèles de distribution d’allèles pour la détection de la variabilité génétique adaptative chez une espèce non modèle, Arabis alpina". Grenoble, 2010. http://www.theses.fr/2010GRENV034.
Texto completo da fonteUnderstanding the molecular basis of adaptation is a major task in evolutionary biology. Local adaptation is the pattern of genotype distributions driven by the natural selection that tends to differentiate populations living in different environments. Genetically, local adaptation results in allele frequencies varying along selection gradients. Our objective is to infer the contribution of allele distribution models in the study of local adaptation through the case of the alpine plant Arabis alpina (Brassicaceae) in the wild. First, a genome scan of 825 AFLP markers genotyped on 678 plants from 198 sites in French and Swiss Alps has been completed and has required the development of a semi-automatic method to select the markers. The effects of this selection on the estimation of genetic structure and variability have been explored. Second, ecologically relevant loci were identified as potentially submitted to selection. Their allele distributions are significantly correlated with environmental variables and topographical conditions. The confounding effects (admixture and isolation by distance) were assessed and discarded in our study case. Some ecologically relevant loci have been sequenced to identify candidate genes and genomic regions potentially selected using the synteny between the genomes of A. Alpina and the model species Arabidopsis thaliana. Finally, the correlative approach to detect selection was compared with more traditional approaches of population genomic. These results suggest that the allele distribution models are a first step before the relevant functional ecology studies to better understand the adaptation to different environmental conditions
Aynaud, Thomas. "Détection de communautés dans les réseaux dynamiques". Paris 6, 2011. http://www.theses.fr/2011PA066438.
Texto completo da fonteMost complex networks have a particular structure in which nodes are arranged in groups, called communities, with many internal links but only a few between them. The identification of communities gives insights on the structure of the graph and is important in many contexts. We will study this structure in the case of dynamic networks using two different approaches. The first approach consists in tracking communities over time by detecting them at every timestep and following their evolution. We will see that although very natural, this approach raises many questions of stability: the algorithms tend to change their results a lot even if the network changes only a little. This implies that the observed changes in the communities are in fact related to the algorithm and not to real transformations in network structure. We therefore propose an analysis of the instability of three algorithms and a solution to the instability. The second approach consists in detecting the community structure not just for a moment but for a period of time called the time window. The length of the time window is then a crucial problem and we propose a hierachical time segmentation method in time windows. Moreover, the time windows do not have to be contiguous allowing for example to detect a repeating structure. Finally, we conclude with applications to event detection on the Internet and segmentation of videos. We will show that we can detect events by finding the times when the structure changes abruptly. For the segmentation of videos, we also had stability issues and thus we have developed a more stable tracking and detection algorithm
Rio, Maxime. "Modèles bayésiens pour la détection de synchronisations au sein de signaux électro-corticaux". Phd thesis, Université de Lorraine, 2013. http://tel.archives-ouvertes.fr/tel-00859307.
Texto completo da fonteHossen, Karim. "Inférence automatique de modèles d'applications Web et protocoles pour la détection de vulnérabilités". Thesis, Grenoble, 2014. http://www.theses.fr/2014GRENM077/document.
Texto completo da fonteIn the last decade, model-based testing (MBT) approaches have shown their efficiency in the software testing domain but a formal model of the system under test (SUT) is required and, most of the time, not available for several reasons like cost, time or rights. The goal of the SPaCIoS project is to develop a security testing tool using MBT approach. The goal of this work, funded by the SPaCIoS project, is to develop and implement a model inference method for Web applications and protocols. From the inferred model, vulnerability detection can be done following SPaCIoS model-checking method or by methods we have developed. We developed an inference algorithm adapted to Web applications and their properties. This method takes into account application data and their influence on the control flow. Using data mining algorithms, the inferred model is refined with optimized guards and output functions. We also worked on the automation of the inference. In active learning approaches, it is required to know the complete interface of the system in order to communicate with it. As this step can be time-consuming, this step has been made automatic using crawler and interface extraction method optimized for inference. This crawler is also available as standalone for third-party inference tools. In the complete inference algorithm, we have merged the inference algorithm and the interface extraction to build an automatic procedure. We present the free software SIMPA, containing the algorithms, and we show some of the results obtained on SPaCIoS case studies and protocols
Cheynet, Valérie. "De la détection du virus VIH-1 : protéines recombinantes et modèles cellulaires d'infection". Lyon 1, 1994. http://www.theses.fr/1994LYO1T211.
Texto completo da fonteBelard, Nuno. "Raisonnement sur les modèles : détection et isolation d'anomalies dans les systèmes de diagnostic". Toulouse 3, 2012. http://thesesups.ups-tlse.fr/1697/.
Texto completo da fonteIn Model-Based Diagnosis, a set of inference rules is typically used to compute diagnoses using a scientific and mathematical theory about a system under study and some observations. Contrary to the classical hypothesis, it is often the case that these Models are abnormal with respect to a series of required properties, hence affecting the quality of the computed diagnoses with possibly huge economical consequences, in particular at Airbus. A thesis on reality and cognition is firstly used to redefine the classic framework of model-based diagnosis from a formal model-theoretic perspective. This, in turn, enables the formalisation of abnormalities and of their relation with the properties diagnoses. With such material and the idea that an implemented diagnostic system can be seen a real-world artefact to be diagnosed, a theory of meta-diagnosis is developed, enabling the detection and isolation of abnormalities in Models of diagnostic systems and explanation in general. Such theory is then encoded in a tool, called MEDITO, and successfuly tested against Airbus real-world industrial problems. Moreover, as different heterogeneous implemented Airbus diagnostic systems, suffering from distinct abnormalities, may compute different diagnoses, methods and tools are developed for: 1) checking the consistency between subsystem-level diagnoses and 2) validating and comparing the performance of these diagnostic systems. Such work relies on an original bridge between the Airbus framework of diagnosis and its academic counterpart. Finally, meta-diagnosis is generalised to handle meta-systems other than implemented diagnostic systems
Plesse, François. "Intégration de Connaissances aux Modèles Neuronaux pour la Détection de Relations Visuelles Rares". Thesis, Paris Est, 2020. http://www.theses.fr/2020PESC1003.
Texto completo da fonteData shared throughout the world has a major impact on the lives of billions of people. It is critical to be able to analyse this data automatically in order to measure and alter its impact. This analysis is tackled by training deep neural networks, which have reached competitive results in many domains. In this work, we focus on the understanding of daily life images, in particular on the interactions between objects and people that are visible in images, which we call visual relations.To complete this task, neural networks are trained in a supervised manner. This involves minimizing an objective function that quantifies how detected relations differ from annotated ones. Performance of these models thus depends on how widely and accurately annotations cover the space of visual relations.However, existing annotations are not sufficient to train neural networks to detect uncommon relations. Thus we integrate knowledge into neural networks during the training phase. To do this, we model semantic relationships between visual relations. This provides a fuzzy set of relations that more accurately represents visible relations. Using the semantic similarities between relations, the model is able to learn to detect uncommon relations from similar and more common ones. However, the improved training does not always translate to improved detections, because the objective function does not capture the whole relation detection process. Thus during the inference phase, we combine knowledge to model predictions in order to predict more relevant relations, aiming to imitate the behaviour of human observers
Ginolhac, Guillaume. "Utilisation de modèles de réverbération pour améliorer la détection en acoustique sous-marine". Grenoble INPG, 2001. http://www.theses.fr/2001INPG0112.
Texto completo da fonteSaidi, Yacine. "Méthodes appliquées de détection et d'estimation de rupture dans les modèles de régression". Université Joseph Fourier (Grenoble ; 1971-2015), 1986. http://tel.archives-ouvertes.fr/tel-00319930.
Texto completo da fonteRahal, Mohamed Ilyas. "Génération d'algorithmes de diagnostic robustes à base de modèles bond graph hybrides". Thesis, Lille 1, 2016. http://www.theses.fr/2016LIL10029/document.
Texto completo da fonteThe present PH.D thesis deals with integrated design of robust Fault Detection and Isolation system (FDI) based on Hybrid Bond Graph (HBG) in Linear Fractional Transformation (LFT) form. Based on consulted literature about hybrid systems, each operating mode is mainly modelled by specific model for which are generated determinist fault indicators. The innovative interest of developed research can be summarized as follows: (1) use only one HBG uncertain model based on controlled junctions and representing all operating modes, (2) structural and causal properties of the LFT HBG are exploited for systematic generation of Global Analytical Redundancy Relations (GARRs), and detection thresholds, robust to parameter uncertainties, and (3) finally use of only one tool: the Diagnosis Hybrid Bond Graph (DHBG) for not only modelling but also for online surveillance. The developed approach is illustrated by electrical circuit pedagogical example and application to hydraulic system
Boursier, Yannick. "Ejections coronales de masse : détection, propriétés statistiques et reconstruction 3D". Aix-Marseille 3, 2007. http://www.theses.fr/2007AIX30082.
Texto completo da fonteGabard, Christophe. "Détection et suivi de cibles dans un environnement non-contraint". Paris 6, 2013. http://www.theses.fr/2013PA066536.
Texto completo da fonteIntelligent video surveillance systems are often complicated to set up and depend on parameters which are difficult to control. The main goal of the proposed thesis is to reduce the complexity to use these video surveillance systems and to propose generic methods. At first, with a target model, we fully model the segmentation process in order to estimate a priori statistics on the results. By reversing these theoretical expressions, we are thus able to find the threshold to be used to achieve a desired statistical detection rate, thus providing an automatic parameterization method. We then propose a detection algorithm that uses a modeling of the whole scene called SMOG is based on a single and global mixture of Gaussian distributions for both the background and the targets. A mode, describing a group of pixels in a 5D space (color and position), characterizes as well the appearance and the shape of the pixels group. The proposed approach combines the accuracy of pixel-wise decision to the robustness of a decision based on a group of pixels and provides a first level of object tracking. Based on information provided by the detection algorithm, a generic algorithm for multi-target tracking has been proposed. The method generates different hypotheses, each one representing a possible evolution of different tracks. The likelihood of each hypothesis is estimated and used to retain only the most relevant ones. To obtain a generic algorithm a large part of the thresholds are automatically estimated from the generated models
Diallo, Boubacar. "Mesure de l'intégrité d'une image : des modèles physiques aux modèles d'apprentissage profond". Thesis, Poitiers, 2020. http://www.theses.fr/2020POIT2293.
Texto completo da fonteDigital images have become a powerful and effective visual communication tool for delivering messages, diffusing ideas, and proving facts. The smartphone emergence with a wide variety of brands and models facilitates the creation of new visual content and its dissemination in social networks and image sharing platforms. Related to this phenomenon and helped by the availability and ease of use of image manipulation softwares, many issues have arisen ranging from the distribution of illegal content to copyright infringement. The reliability of digital images is questioned for common or expert users such as court or police investigators. A well known phenomenon and widespread examples are the "fake news" which oftenly include malicious use of digital images.Many researchers in the field of image forensic have taken up the scientific challenges associated with image manipulation. Many methods with interesting performances have been developed based on automatic image processing and more recently the adoption of deep learning. Despite the variety of techniques offered, performance are bound to specific conditions and remains vulnerable to relatively simple malicious attacks. Indeed, the images collected on the Internet impose many constraints on algorithms questioning many existing integrity verification techniques. There are two main peculiarities to be taken into account for the detection of a falsification: one is the lack of information on pristine image acquisition, the other is the high probability of automatic transformations linked to the image-sharing platforms such as lossy compression or resizing.In this thesis, we focus on several of these image forensic challenges including camera model identification and image tampering detection. After reviewing the state of the art in the field, we propose a first data-driven method for identifying camera models. We use deep learning techniques based on convolutional neural networks (CNNs) and develop a learning strategy considering the quality of the input data versus the applied transformation. A family of CNN networks has been designed to learn the characteristics of the camera model directly from a collection of images undergoing the same transformations as those commonly used on the Internet. Our interest focused on lossy compression for our experiments, because it is the most used type of post-processing on the Internet. The proposed approach, therefore, provides a robust solution to compression for camera model identification. The performance achieved by our camera model detection approach is also used and adapted for image tampering detection and localization. The performances obtained underline the robustness of our proposals for camera model identification and image forgery detection
Lyazrhi, Faouzi. "Procédures optimales de détection de ruptures dans un modèle linéaire gaussien". Toulouse 3, 1993. http://www.theses.fr/1993TOU30076.
Texto completo da fonteSoullard, Yann. "Classification et détection de figures chartistes par apprentissage statistique". Paris 6, 2013. http://www.theses.fr/2013PA066341.
Texto completo da fonteThis thesis deals with financial stock market analysis and is especially focused on chart pattern recognition. A chart pattern is a particular shape which has a predictive power; it is defined by theoretical rules. Detecting such patterns is difficult. There is an important gap between theory and practice; real patterns do not perfectly respect the theoretical rules. Moreover, chart patterns definition seems subjective; it depends on the financial expert. Finally, there is no large labeled datasets of chart patterns. We study classification and detection of chart patterns using statistical markovian systems. We focus on generative (Hidden Markov Models) and discriminative (Conditional Random Fields, Hidden CRFs) approaches which are standard technologies for sequential data recognition. We propose various strategies to learn accurate systems with small training sets. The first one blends HMMs and HCRFs in such a way that the modeling ability of the generative models is used to limit the overfitting of the discriminative ones. The second strategy, is a semi-supervised approach which learns jointly a HMM and a HCRF systems; it has some similarity with the well-known co-training algorithm. To design an accurate detection system dedicated to a particular financial expert, we propose a two level system where candidate patterns are first extracted from the financial stock-market using HMMs, and then they are confirmed as chart patterns or rejected by a SVM which uses an enriched representation of patterns. While the HMM system is learn once for every expert, the SVM level is trained with an active learning strategy to take into account the expert’s own detection criteria
Belbachir, Faiza. "Approches basées sur les modèles de langue pour la recherche d'opinions". Toulouse 3, 2014. http://thesesups.ups-tlse.fr/2341/.
Texto completo da fonteEvolution of the World Wide Web has brought us various forms of data like factual data, product reviews, arguments, discussions, news data, temporal data, blog data etc. The blogs are considered to be the best way for the expression of the one's opinions about something including from a political subject to a product. These opinions become more important when they influence govt. Policies or companies marketing agendas and much more because of their huge presence on the web. Therefore, it becomes equally important to have such information systems that could process this kind of information on the web. In this thesis, we propose approach (es) that distinguish between factual and opinion documents with the purpose of further processing of opinionated information. Most of the current opinion finding approaches, some base themselves on lexicons of subjective terms while others exploit machine learning techniques. Within the framework of this thesis, we are interested in both types of approaches by mitigating some of their limits. Our contribution revolves around three main aspects of opinion mining. First of all we propose a lexical approach for opinion finding task. We exploit various subjective publicly available resources such as IMDB, ROTTEN, CHESLY and MPQA that are considered to be opinionated data collections. The idea is that if a document is similar to these, it is most likely that it is an opinionated document. We seek support of language modeling techniques for this purpose. We model the test document (i. E. The document whose subjectivity is to be evaluated) the source of opinion by language modeling technique and measure the similarity between both models. The higher the score of similarity is, the more the document subjective. Our second contribution of detection of opinion is based on the machine learning. For that purpose, we propose and evaluate various features such as the emotivity, the subjectivity, the addressing, the reflexivity and report results to compare them with current approaches. Our third contribution concerns the polarity of the opinion which determines if a subjective document has a positive or negative opinion on a given topic. We conclude that the polarity of a term can depend on the domain in which it is being used
Dib, Linda. "Détection des mutations simultanées dans les séquences protéiques non-divergentes". Paris 6, 2012. http://www.theses.fr/2012PA066016.
Texto completo da fonteOk, David. "Mise en correspondance robuste et détection de modèles visuels appliquées à l'analyse de façades". Phd thesis, Université Paris-Est, 2013. http://pastel.archives-ouvertes.fr/pastel-00974556.
Texto completo da fonteOk, David, e David Ok. "Mise en correspondance robuste et détection de modèles visuels appliquées à l'analyse de façades". Phd thesis, Université Paris-Est, 2013. http://pastel.archives-ouvertes.fr/tel-00844049.
Texto completo da fonteCoulon, Martial. "Contribution à la détection de modèles paramétriques en présence de bruit additif et multiplicatif". Toulouse, INPT, 1999. http://www.theses.fr/1999INPT035H.
Texto completo da fonteAltuve, Miguel. "Détection multivariée des épisodes d'apnée-bradycardie chez le prématuré par modèles semi-markovien cachés". Rennes 1, 2011. http://www.theses.fr/2011REN1S053.
Texto completo da fonteThis dissertation studies the early detection of apnea-bradycardia (AB) events in preterm infants. After defining the importance of AB detection from a clinical point of view, a methodological approach is proposed. It relies on a data mining process that includes data cleansing and feature extraction. In chapter 3, a novel method based on evolutionary algorithms, for optimizing the thresholds and the analysis windows, is proposed to adapt the algorithms of the ECG signal to the specific characteristics of preterm infants, very different from the EGC of adult. In chapter 4, a semi-Markovian approach is adapted for modeling of dynamics and several improvements are proposed : heterogeneous models, adaptation to online processing, optimization of experiments, are reported on simulated and read signals. They clearly highlight the importance of considering the dynamic of the signals. They also emphasize that with a suitable pre-treatment such as the quantification of observations and the introduction of delay between the observable, a significant gain in performance can be observed
Rusch, Philippe. "Modèles d'écoulement de globules rouges à travers un réseau capillaire : détection d'effets non linéaires". Université Louis Pasteur (Strasbourg) (1971-2008), 1989. http://www.theses.fr/1989STR13203.
Texto completo da fonteIzard, Camille. "Modélisation et estimation statistique pour l'imagerie médicale : application à la détection d'amers". Thesis, Lille 1, 2008. http://www.theses.fr/2008LIL10026/document.
Texto completo da fonteWe present a family of statistical mode/s based on deformable template for medical image analysis, and more specifically for the detection of anatomical landmarks. Deformable template models are commonly used for image matching to perform segmentation, registration or classification. We show that if the position of the landmarks characterizes uniquely the deformation of an image, the landmark detection problem can be formalized as a local matching problem. Based on the proposed statistical models and using maximum Iikelihood principles, we derive both an algorithm to learn the model from training data and a testing algorithm for the detection of landmarks in new images. The first two statistical models we propose rely on intensity or edge matching to identify the location of the landmarks; while the third one uses simultaneous image segmentation and template registration to locate the landmarks. We introduce a foregroundlbackground statistical model for medical imaging, which allows us to limit the computational effort to matching discriminative patterns surrounding the land marks. The proposed a/gorithms provide simple generic methods to perform automatic detection of landmarks in medical imaging. We tested our approach on the detection of landmarks ln brain Magnetic Resonance Images
Demattei, Christophe. "Détection d'agrégats temporels et spatiaux". Phd thesis, Université Montpellier I, 2006. http://tel.archives-ouvertes.fr/tel-00134491.
Texto completo da fonteNous proposons une revue des méthodes existantes ainsi que notre contribution dans différentes directions. Deux approches sont proposées dans le cadre temporel permettant pour l'une d'éviter l'utilisation de simulations et pour l'autre de prendre en compte les données dont l'information temporelle est incomplète. Nous avons également mis au point une méthode de détection de clusters spatiaux de forme arbitraire permettant d'analyser des données dont on connaît la localisation géographique exacte. Cette approche a été appliquée sur des données particulières, celles obtenues par Imagerie par Résonance Magnétique fonctionnelle. Les perspectives d'analyse spatio-temporelle sont finalement évoquées.
Vaezi-Nejad, Hossein. "Détection de défauts d'instruments de mesure". Nancy 1, 1990. http://docnum.univ-lorraine.fr/public/SCD_T_1990_0016_VAEZI_NEJAD.pdf.
Texto completo da fonteNel, François. "Suivi de mouvements informationnels : construction, modélisation et simulation de graphes de citations, application à la détection de buzz". Paris 6, 2011. http://www.theses.fr/2011PA066541.
Texto completo da fonteChoukri, Karim. "Un formalisme pour les tests statistiques de conformité de modèles pour des séries chronologiques : application à la détection de changements de modèles". Ecole Nationale Supérieure des Télécommunications(Paris), 1994. http://www.theses.fr/1994ENST0027.
Texto completo da fonteLavarde, 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.
Texto completo da fonteThis 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
Haddadi, Souad. "Réseaux de neurones, textures et modèles markoviens pour la détection et l'identification d'objets en mouvement". Compiègne, 1997. http://www.theses.fr/1997COMP1081.
Texto completo da fonteIn this PhD thesis, we present a method of analysis for image sequences. The method aims at dynamic scene interpretation where arbitrary objects evolve (in particular, human beings) and the scenes present non-uniform backgrounds and non-controlled illumination. Two processing approaches have been aborded : movement analysis (moving object detection) and pattern recognition (object identification). The proposed detection approach relies on a statistical segmentation procedure, which is based on the markovian principle and the analysis of texture. Considering an operator based on the differences between three successive images, taken two at a time, moving objects are detected, as well as the background regions which are discovered or occluded by these objects during their displacement. A coarse segmentation of this image operator is then applied to process the relevant zones of the image. This operation is then linked to a finer segmentation based on the markovian and textural principle. This problem was approached to a classification of the image operator into fixed and moving pixels. The identification approach of these objects uses another type of statistical model : the artificial neural networks, which allow computer training, after examples. Thus, models of neural network architectures were developed and applied to human being identification. The performances of these networks were calculated using two databases built for this project. We have demonstrated that high performances could be attained using MLP-type networks for our application. However, the studies accomplished during this thesis reveal a certain number of difficult problems. For example, in several cases we confronted the problem of selecting a pertinent training set
Zheng, Huicheng. "Modèles de maximum d'entropie pour la détection de la peau : application au filtrage de l'internet". Lille 1, 2004. https://pepite-depot.univ-lille.fr/LIBRE/Th_Num/2004/50376-2004-Zheng.pdf.
Texto completo da fonteDans un troisième temps, nous ajoutons des contraintes sur les couleurs des pixels voisins. Le modèle obtenu est à nouveau un champs de Markov mais contenant un très grand nombre de paramètres. Afin, aussi bien d'estimer les paramètres de ces modèles que d'effectuer de l'inférence, c'est à dire, étant donné une image couleur, calculer la probabilité pour chaque pixel qu'il corresponde à de la peau, nous proposons d'approximer, localement, le graphe associé aux pixels par un arbre. On dispose alors de l'algorithme "iterative scaling" pour l'estimation des paramètres et de l'algorithme "belief propagation" pour l'inférence. Nous avons effectué de nombreuses études expérimentales afin d'évaluer les performances respectives des différents modèles, en particulier en modifiant la taille et la géométrie des arbres. Dans le cas du projet européen Poesia, nous avons utilisé notre détecteur de peau en entrée d'un système de classification utlisant la méthode des réseaux neuronaux pour bloquer les pages webs indésirable pour les enfants. Nous avons obtenu des résultats extrèmement encourageants
Parisot, Sarah. "Compréhension, modélisation et détection de tumeurs cérébrales : modèles graphiques et méthodes de recalage/segmentation simultanés". Phd thesis, Ecole Centrale Paris, 2013. http://tel.archives-ouvertes.fr/tel-00944541.
Texto completo da fonteLiebelt, Jörg. "Détection de classes d'objets et estimation de leurs poses à partir de modèles 3D synthétiques". Grenoble, 2010. https://theses.hal.science/tel-00553343.
Texto completo da fonteThis dissertation aims at extending object class detection and pose estimation tasks on single 2D images by a 3D model-based approach. The work describes learning, detection and estimation steps adapted to the use of synthetically rendered data with known 3D geometry. Most existing approaches recognize object classes for a particular viewpoint or combine classifiers for a few discrete views. By using existing CAD models and rendering techniques from the domain of computer graphics which are parameterized to reproduce some variations commonly found in real images, we propose instead to build 3D representations of object classes which allow to handle viewpoint changes and intra-class variability. These 3D representations are derived in two different ways : either as an unsupervised filtering process of pose and class discriminant local features on purely synthetic training data, or as a part model which discriminatively learns the object class appearance from an annotated database of real images and builds a generative representation of 3D geometry from a database of synthetic CAD models. During detection, we introduce a 3D voting scheme which reinforces geometric coherence by means of a robust pose estimation, and we propose an alternative probabilistic pose estimation method which evaluates the likelihood of groups of 2D part detections with respect to a full 3D geometry. Both detection methods yield approximate 3D bounding boxes in addition to 2D localizations ; these initializations are subsequently improved by a registration scheme aligning arbitrary 3D models to optical and Synthetic Aperture Radar (SAR) images in order to disambiguate and prune 2D detections and to handle occlusions. The work is evaluated on several standard benchmark datasets and it is shown to achieve state-of-the-art performance for 2D detection in addition to providing 3D pose estimations from single images
Harrou, Fouzi. "Détection d'anomalies en présence de paramètres de nuisance bornés". Troyes, 2010. http://www.theses.fr/2010TROY0002.
Texto completo da fonteAnomaly detection is addressed within a statistical framework. Often the statistical model is composed of two types of parameters: the informative parameters and the nuisance ones. The nuisance parameters are of no interest for detection but they are necessary to complete the model. In the case of unknown, non-random and non-bounded nuisance parameters, their elimination is unavoidable. Unfortunately, this can lead to a serious degradation of the detector capacity because some anomalies are masked by nuisance parameters. Nevertheless, in many cases, the physical nature of nuisance parameter is known, and this may allow set bounds to the values taken by this parameter. In this work, the problem of anomaly detection with bounded nuisance parameters has been addressed from the statistical point of view in the context of linear model. The con-strained generalized likelihood ratio test has been studied. It has been shown that the performances of anomaly detector can be drastically improved by taking into account the lower and upper bounds, naturally imposed on the nuisance parameters. Some applications to integrity control of GPS positioning systems are developed in fields of train navigation. Finally, the detection of abnormal ozone measurements by using a regional ozone surveillance network has been used to illustrate the theoretical findings and to show the relevance of the proposed method
Louis, Huguette. "Détection de l'activation des CML vasculaires : implications dans la différenciation : approches dans les modèles animaux et in vitro : double détection immunologique/hybridation in situ froide". Bordeaux 2, 1998. http://www.theses.fr/1998BOR28605.
Texto completo da fonteMorvidone, Marcela. "Etude et comparaison d'algorithmes de détection optimale pour les signaux modulés en amplitude et en fréquence : applications aux ondes gravitationnelles". Aix-Marseille 1, 2002. http://www.theses.fr/2002AIX11063.
Texto completo da fonte