Дисертації з теми "Apprentissage parcimonieux"
Оформте джерело за APA, MLA, Chicago, Harvard та іншими стилями
Ознайомтеся з топ-50 дисертацій для дослідження на тему "Apprentissage parcimonieux".
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.
Avalos, Marta. "Modèles additifs parcimonieux." Phd thesis, Université de Technologie de Compiègne, 2004. http://tel.archives-ouvertes.fr/tel-00008802.
Повний текст джерелаAmate, Laure. "Apprentissage de modèles de formes parcimonieux basés sur des représentations splines." Phd thesis, Université de Nice Sophia-Antipolis, 2009. http://tel.archives-ouvertes.fr/tel-00456612.
Повний текст джерелаAmate, Laure. "Apprentissage de modèles de formes parcimonieux basés sur les représentations splines." Nice, 2009. http://www.theses.fr/2009NICE4117.
Повний текст джерелаIn many contexts it is important to be able to find compact representations of the collective morphological properties of a set of objects. This is the case of autonomous robotic platforms operating in natural environments that must use the perceptual properties of the objects present in their workspace to execute their mission. This thesis is a contribution to the definition of formalisms and methods for automatic identification of such models. The shapes we want to characterize are closed curves corresponding to contours of objects detected in the scene. We begin with the formal definition of the notion of shape as classes of equivalence with respect to groups of basic geometric operators, introducing two distinct approaches that have been used in the literature: discrete and continuous. The discrete theory, admitting the existence of a finite number of recognizable landmarks, provides in an obvious manner a compact representation but is sensible to their selection. The continuous theory of shapes provides a more fundamental approach, but leads to shape spaces of infinite dimension, lacking the parsimony of the discrete representation. We thus combine in our work the advantages of both approaches representing shapes of curves with splines: piece-wise continuous polynomials defined by sets of knots and control points. We first study the problem of fitting free-knots splines of varying complexity to a single observed curve. The trade-o_ between the parsimony of the representation and its fidelity to the observations is a well known characteristic of model identification using nested families of increasing dimension. After presenting an overview of methods previously proposed in the literature, we single out a two-step approach which is formally sound and matches our specific requirements. It splits the identification, simulating a reversible jump Markov chain to select the complexity of the model followed by a simulated annealing algorithm to estimate its parameters. We investigate the link between Kendall's shape space and spline representations when we take the spline control points as landmarks. We consider now the more complex problem of modeling a set of objects with similar morphological characteristics. We equate the problem to finding the statistical distribution of the parameters of the spline representation, modeling the knots and control points as unobserved variables. The identified distribution is the maximizer of a marginal likelihood criterion, and we propose a new Expectation-Maximization algorithm to optimize it. Because we may want to treat a large number of curves observed sequentially, we adapt an iterative (on-line) version of the EM algorithm recently proposed in the literature. For the choice of statistical distributions that we consider, both the expectation and the maximization steps must resort to numerical approximations, leading to a stochastic/on-line variant of the EM algorithm that, as far as we know, is implemented here for the first time
Huet, Romain. "Codage neural parcimonieux pour un système de vision." Thesis, Lorient, 2017. http://www.theses.fr/2017LORIS439/document.
Повний текст джерелаThe neural networks have gained a renewed interest through the deep learning paradigm. Whilethe so called optimised neural nets, by optimising the parameters necessary for learning, require massive computational resources, we focus here on neural nets designed as addressable content memories, or neural associative memories. The challenge consists in realising operations, traditionally obtained through computation, exclusively with neural memory in order to limit the need in computational resources. In this thesis, we study an associative memory based on cliques, whose sparse neural coding optimises the data diversity encoded in the network. This large diversity allows the clique based network to be more efficient in messages retrieval from its memory than other neural associative memories. The associative memories are known for their incapacity to identify without ambiguities the messages stored in a saturated memory. Indeed, depending of the information present in the network and its encoding, a memory can fail to retrieve a desired result. We are interested in tackle this issue and propose several contributions in order to reduce the ambiguities in the cliques based neural network. Besides, these cliques based nets are unable to retrieve an information within their memories if the message is unknown. We propose a solution to this problem through a new associative memory based on cliques which preserves the initial network's corrective ability while being able to hierarchise the information. The hierarchy relies on a surjective and bidirectional transition to generalise an unknown input with an approximation of learnt information. The associative memories' experimental validation is usually based on low dimension artificial dataset. In the computer vision context, we report here the results obtained with real datasets used in the state-of-the-art, such as MNIST, Yale or CIFAR
Belilovsky, Eugene. "Apprentissage de graphes structuré et parcimonieux dans des données de haute dimension avec applications à l’imagerie cérébrale." Thesis, Université Paris-Saclay (ComUE), 2018. http://www.theses.fr/2018SACLC027.
Повний текст джерелаThis dissertation presents novel structured sparse learning methods on graphs that address commonly found problems in the analysis of neuroimaging data as well as other high dimensional data with few samples. The first part of the thesis proposes convex relaxations of discrete and combinatorial penalties involving sparsity and bounded total variation on a graph as well as bounded `2 norm. These are developed with the aim of learning an interpretable predictive linear model and we demonstrate their effectiveness on neuroimaging data as well as a sparse image recovery problem.The subsequent parts of the thesis considers structure discovery of undirected graphical models from few observational data. In particular we focus on invoking sparsity and other structured assumptions in Gaussian Graphical Models (GGMs). To this end we make two contributions. We show an approach to identify differences in Gaussian Graphical Models (GGMs) known to have similar structure. We derive the distribution of parameter differences under a joint penalty when parameters are known to be sparse in the difference. We then show how this approach can be used to obtain confidence intervals on edge differences in GGMs. We then introduce a novel learning based approach to the problem structure discovery of undirected graphical models from observational data. We demonstrate how neural networks can be used to learn effective estimators for this problem. This is empirically shown to be flexible and efficient alternatives to existing techniques
Mattei, 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
Hérault, Romain. "Vision et apprentissage statistique pour la reconnaissance d'items comportementaux." Compiègne, 2007. http://www.theses.fr/2007COMP1715.
Повний текст джерелаThis work consists in the detection of behavioral items in order to prevent driver drowsiness. Videos were shot from within a car, and each picture of the video was characterized by six behavioral items. Our work consists in the retrieval of these items, picture by picture. The study was decomposed into two phases: 1) A Head and facial action tracking. A framework to 3D head pose and facial action tracking with an adaptive appearance model based on a mixture model is proposed to deal with face occlusion ; 2) A recognition of the behavioral items based on data retrieved from the tracking. We propose a new criterion leading to an adaptation of maximum likelihood estimation. The model outputs proper conditional probabilities into a user-defined interval. This criterion is applied to MLPs and IOHMMs for the recognition of the behavioral items
Meghnoudj, Houssem. "Génération de caractéristiques à partir de séries temporelles physiologiques basée sur le contrôle optimal parcimonieux : application au diagnostic de maladies et de troubles humains." Electronic Thesis or Diss., Université Grenoble Alpes, 2024. http://www.theses.fr/2024GRALT003.
Повний текст джерелаIn this thesis, a novel methodology for features generation from physiological signals (EEG, ECG) has been proposed that is used for the diagnosis of a variety of brain and heart diseases. Based on sparse optimal control, the generation of Sparse Dynamical Features (SDFs) is inspired by the functioning of the brain. The method's fundamental concept revolves around sparsely decomposing the signal into dynamical modes that can be switched on and off at the appropriate time instants with the appropriate amplitudes. This decomposition provides a new point of view on the data which gives access to informative features that are faithful to the brain functioning. Nevertheless, the method remains generic and versatile as it can be applied to a wide range of signals. The methodology's performance was evaluated on three use cases using openly accessible real-world data: (1) Parkinson's Disease, (2) Schizophrenia, and (3) various cardiac diseases. For all three applications, the results are highly conclusive, achieving results that are comparable to the state-of-the-art methods while using only few features (one or two for brain applications) and a simple linear classifier supporting the significance and reliability of the findings. It's worth highlighting that special attention has been given to achieving significant and meaningful results with an underlying explainability
Laporte, Léa. "La sélection de variables en apprentissage d'ordonnancement pour la recherche d'information : vers une approche contextuelle." Toulouse 3, 2013. http://thesesups.ups-tlse.fr/2170/.
Повний текст джерелаLearning-to-rank aims at automatically optimizing a ranking function learned on training data by a machine learning algorithm. Existing approaches have two major drawbacks. Firstly, the ranking functions can use several thousands of features, which is an issue since algorithms have to deal with large scale data. This can also have a negative impact on the ranking quality. Secondly, algorithms learn an unique fonction for all queries. Then, nor the kind of user need neither the context of the query are taken into account in the ranking process. Our works focus on solving the large-scale issue and the context-aware issue by using feature selection methods dedicated to learning-to-rank. We propose five feature selection algorithms based on sparse Support Vector Machines (SVM). Three proceed to feature selection by reweighting the L2-norm, one solves a L1-regularized problem whereas the last algorithm consider nonconvex regularizations. Our methods are faster and sparser than state-of-the-art algorithms on benchmark datasets, while providing similar performances in terms of RI measures. We also evaluate our approches on a commercial dataset. Experimentations confirm the previous results. We propose in this context a relevance model based on users clicks, in the special case of multi-clickable documents. Finally, we propose an adaptative and query-dependent ranking system based on feature selection. This system considers several clusters of queries, each group defines a context. For each cluster, the system selects a group of features to learn a context-aware ranking function
Vezard, Laurent. "Réduction de dimension en apprentissage supervisé : applications à l’étude de l’activité cérébrale." Thesis, Bordeaux 1, 2013. http://www.theses.fr/2013BOR15005/document.
Повний текст джерелаThe aim of this work is to develop a method able to automatically determine the alertness state of humans. Such a task is relevant to diverse domains, where a person is expected or required to be in a particular state. For instance, pilots, security personnel or medical personnel are expected to be in a highly alert state, and this method could help to confirm this or detect possible problems. In this work, electroencephalographic data (EEG) of 58 subjects in two distinct vigilance states (state of high and low alertness) were collected via a cap with $58$ electrodes. Thus, a binary classification problem is considered. In order to use of this work on a real-world applications, it is necessary to build a prediction method that requires only a small number of sensors (electrodes) in order to minimize the time needed by the cap installation and the cap cost. During this thesis, several approaches have been developed. A first approach involves use of a pre-processing method for EEG signals based on the use of a discrete wavelet decomposition in order to extract the energy of each frequency in the signal. Then, a linear regression is performed on the energies of some of these frequencies and the slope of this regression is retained. A genetic algorithm (GA) is used to optimize the selection of frequencies on which the regression is performed. Moreover, the GA is used to select a single electrode .A second approach is based on the use of the Common Spatial Pattern method (CSP). This method allows to define linear combinations of the original variables to obtain useful synthetic signals for the task classification. In this work, a GA and a sequential search method have been proposed to select a subset of electrode which are keep in the CSP calculation.Finally, a sparse CSP algorithm, based on the use of existing work in the sparse principal component analysis, was developed.The results of the different approaches are detailed and compared. This work allows us to obtaining a reliable model to obtain fast prediction of the alertness of a new individual
Vezard, Laurent. "Réduction de dimension en apprentissage supervisé : applications à l'étude de l'activité cérébrale." Phd thesis, Université Sciences et Technologies - Bordeaux I, 2013. http://tel.archives-ouvertes.fr/tel-00944790.
Повний текст джерелаMoreau, Thomas. "Représentations Convolutives Parcimonieuses -- application aux signaux physiologiques et interpétabilité de l'apprentissage profond." Thesis, Université Paris-Saclay (ComUE), 2017. http://www.theses.fr/2017SACLN054/document.
Повний текст джерелаConvolutional representations extract recurrent patterns which lead to the discovery of local structures in a set of signals. They are well suited to analyze physiological signals which requires interpretable representations in order to understand the relevant information. Moreover, these representations can be linked to deep learning models, as a way to bring interpretability intheir internal representations. In this disserta tion, we describe recent advances on both computational and theoretical aspects of these models.First, we show that the Singular Spectrum Analysis can be used to compute convolutional representations. This representation is dense and we describe an automatized procedure to improve its interpretability. Also, we propose an asynchronous algorithm, called DICOD, based on greedy coordinate descent, to solve convolutional sparse coding for long signals. Our algorithm has super-linear acceleration.In a second part, we focus on the link between representations and neural networks. An extra training step for deep learning, called post-training, is introduced to boost the performances of the trained network by making sure the last layer is optimal. Then, we study the mechanisms which allow to accelerate sparse coding algorithms with neural networks. We show that it is linked to afactorization of the Gram matrix of the dictionary.Finally, we illustrate the relevance of convolutional representations for physiological signals. Convolutional dictionary learning is used to summarize human walk signals and Singular Spectrum Analysis is used to remove the gaze movement in young infant’s oculometric recordings
Lafargue, Raphaël. "Few-shot learning, a data-centric approach for adaptation." Electronic Thesis or Diss., Ecole nationale supérieure Mines-Télécom Atlantique Bretagne Pays de la Loire, 2024. http://www.theses.fr/2024IMTA0439.
Повний текст джерелаThis thesis presents three key contributions aimed at advancing Few-Shot Learning (FSL) through improved model robustness, accurate performance assessment, and task-specific adaptation. First, we explore methods for building robust feature extractors by incorporating simple ingredients, achieving state-of-the-art performance in in-domain classification tasks. Next, we address the need for reliable evaluations of FSL methods by showing that the predominent evaluation protocol is misleading in that it does not account for the randomness of the data, leading to conclusions that may be dataset specific. We propose evaluation techniques that account for this randomness, demonstrating that claims of superiority between methods can change under these considerations. Lastly, we introduce a data-centric approach that enhances cross-domain task adaptation by selectively forgetting portions of the pretraining dataset, reallocating feature space to improve generalization. Together, these contributions provide comprehensive insights for developing robust, adaptable FSL models
Liu, Yuan. "Représentation parcimonieuse basée sur la norme ℓ₀ Mixed integer programming for sparse coding : application to image denoising Incoherent dictionary learning via mixed-integer programming and hybrid augmented Lagrangian". Thesis, Normandie, 2019. http://www.theses.fr/2019NORMIR22.
Повний текст джерелаIn this monograph, we study the exact ℓ₀ based sparse representation problem. For the classical dictionary learning problem, the solution is obtained by iteratively processing two steps: sparse coding and dictionary updating. However, even the problem associated with sparse coding is non-convex and NP-hard. The method for solving this is to reformulate the problem as mixed integer quadratic programming (MIQP). Then by introducing two optimization techniques, initialization by proximal method and relaxation with augmented contraints, the algorithmis greatly speed up (which is thus called AcMIQP) and applied in image denoising, which shows the good performance. Moreover, the classical problem is extended to learn an incoherent dictionary. For dealing with this problem, AcMIQP or proximal method is used for sparse coding. As for dictionary updating, augmented Lagrangian method (ADMM) and extended proximal alternating linearized minimizing method are combined. This exact ℓ₀ based incoherent dictionary learning is applied in image recovery, which illustrates the improved performance with a lower coherence
Vezard, Laurent. "Réduction de dimension en apprentissage supervisé. Application à l'étude de l'activité cérébrale." Phd thesis, Université Sciences et Technologies - Bordeaux I, 2013. http://tel.archives-ouvertes.fr/tel-00926845.
Повний текст джерелаChabiron, Olivier. "Apprentissage d'arbres de convolutions pour la représentation parcimonieuse." Thesis, Toulouse 3, 2015. http://www.theses.fr/2015TOU30213/document.
Повний текст джерелаThe dictionary learning problem has received increasing attention for the last ten years. DL is an adaptive approach for sparse data representation. Many state-of-the-art DL methods provide good performances for problems such as approximation, denoising and inverse problems. However, their numerical complexity restricts their use to small image patches. Thus, dictionary learning does not capture large features and is not a viable option for many applications handling large images, such as those encountered in remote sensing. In this thesis, we propose and study a new model for dictionary learning, combining convolutional sparse coding and dictionaries defined by convolutional tree structures. The aim of this model is to provide efficient algorithms for large images, avoiding the decomposition of these images into patches. In the first part, we study the optimization of a composition of convolutions with sparse kernels, to reach a target atom (such as a cosine, wavelet or curvelet). This is a non-convex matrix factorization problem. We propose a resolution method based on a Gaus-Seidel scheme, which produces good approximations of target atoms and whose complexity is linear with respect to the image size. Moreover, numerical experiments show that it is possible to find a global minimum. In the second part, we introduce a dictionary structure based on convolutional trees. We propose a dictionary update algorithm adapted to this structure and which complexity remains linear with respect to the image size. Finally, a sparse coding step is added to the algorithm in the last part. For each evolution of the proposed method, we illustrate its approximation abilities with numerical experiments
Lesage, Sylvain. "Apprentissage de dictionnaires structurés pour la modélisation parcimonieuse des signaux multicanaux." Phd thesis, Université Rennes 1, 2007. http://tel.archives-ouvertes.fr/tel-00564061.
Повний текст джерелаLesage, Sylvain Bimbot Frédéric. "Apprentissage de dictionnaires structurés pour la modélisation parcimonieuse des signaux multicanaux." [S.l.] : [s.n.], 2007. ftp://ftp.irisa.fr/techreports/theses/2007/lesage.pdf.
Повний текст джерелаBartcus, Marius. "Bayesian non-parametric parsimonious mixtures for model-based clustering." Thesis, Toulon, 2015. http://www.theses.fr/2015TOUL0010/document.
Повний текст джерелаThis thesis focuses on statistical learning and multi-dimensional data analysis. It particularly focuses on unsupervised learning of generative models for model-based clustering. We study the Gaussians mixture models, in the context of maximum likelihood estimation via the EM algorithm, as well as in the Bayesian estimation context by maximum a posteriori via Markov Chain Monte Carlo (MCMC) sampling techniques. We mainly consider the parsimonious mixture models which are based on a spectral decomposition of the covariance matrix and provide a flexible framework particularly for the analysis of high-dimensional data. Then, we investigate non-parametric Bayesian mixtures which are based on general flexible processes such as the Dirichlet process and the Chinese Restaurant Process. This non-parametric model formulation is relevant for both learning the model, as well for dealing with the issue of model selection. We propose new Bayesian non-parametric parsimonious mixtures and derive a MCMC sampling technique where the mixture model and the number of mixture components are simultaneously learned from the data. The selection of the model structure is performed by using Bayes Factors. These models, by their non-parametric and sparse formulation, are useful for the analysis of large data sets when the number of classes is undetermined and increases with the data, and when the dimension is high. The models are validated on simulated data and standard real data sets. Then, they are applied to a real difficult problem of automatic structuring of complex bioacoustic data issued from whale song signals. Finally, we open Markovian perspectives via hierarchical Dirichlet processes hidden Markov models
Hitziger, Sebastian. "Modélisation de la variabilité de l'activité électrique dans le cerveau." Thesis, Nice, 2015. http://www.theses.fr/2015NICE4015/document.
Повний текст джерелаThis thesis investigates the analysis of brain electrical activity. An important challenge is the presence of large variability in neuroelectrical recordings, both across different subjects and within a single subject, for example, across experimental trials. We propose a new method called adaptive waveform learning (AWL). It is general enough to include all types of relevant variability empirically found in neuroelectric recordings, but can be specialized for different concrete settings to prevent from overfitting irrelevant structures in the data. The first part of this work gives an introduction into the electrophysiology of the brain, presents frequently used recording modalities, and describes state-of-the-art methods for neuroelectrical signal processing. The main contribution of this thesis consists in three chapters introducing and evaluating the AWL method. We first provide a general signal decomposition model that explicitly includes different forms of variability across signal components. This model is then specialized for two concrete applications: processing a set of segmented experimental trials and learning repeating structures across a single recorded signal. Two algorithms are developed to solve these models. Their efficient implementation based on alternate minimization and sparse coding techniques allows the processing of large datasets. The proposed algorithms are evaluated on both synthetic data and real data containing epileptiform spikes. Their performances are compared to those of PCA, ICA, and template matching for spike detection
Jas, Mainak. "Contributions pour l'analyse automatique de signaux neuronaux." Electronic Thesis or Diss., Paris, ENST, 2018. http://www.theses.fr/2018ENST0021.
Повний текст джерелаElectrophysiology experiments has for long relied upon small cohorts of subjects to uncover statistically significant effects of interest. However, the low sample size translates into a low power which leads to a high false discovery rate, and hence a low rate of reproducibility. To address this issue means solving two related problems: first, how do we facilitate data sharing and reusability to build large datasets; and second, once big datasets are available, what tools can we build to analyze them ? In the first part of the thesis, we introduce a new data standard for sharing data known as the Brain Imaging Data Structure (BIDS), and its extension MEG-BIDS. Next, we introduce the reader to a typical electrophysiological pipeline analyzed with the MNE software package. We consider the different choices that users have to deal with at each stage of the pipeline and provide standard recommendations. Next, we focus our attention on tools to automate analysis of large datasets. We propose an automated tool to remove segments of data corrupted by artifacts. We develop an outlier detection algorithm based on tuning rejection thresholds. More importantly, we use the HCP data, which is manually annotated, to benchmark our algorithm against existing state-of-the-art methods. Finally, we use convolutional sparse coding to uncover structures in neural time series. We reformulate the existing approach in computer vision as a maximuma posteriori (MAP) inference problem to deal with heavy tailed distributions and high amplitude artifacts. Taken together, this thesis represents an attempt to shift from slow and manual methods of analysis to automated, reproducible analysis
Baccouche, Moez. "Apprentissage neuronal de caractéristiques spatio-temporelles pour la classification automatique de séquences vidéo." Phd thesis, INSA de Lyon, 2013. http://tel.archives-ouvertes.fr/tel-00932662.
Повний текст джерелаPierrefeu, Amicie de. "Apprentissage automatique avec parcimonie structurée : application au phénotypage basé sur la neuroimagerie pour la schizophrénie." Thesis, Université Paris-Saclay (ComUE), 2018. http://www.theses.fr/2018SACLS329/document.
Повний текст джерелаSchizophrenia is a disabling chronic mental disorder characterized by various symptoms such as hallucinations, delusions as well as impairments in high-order cognitive functions. Over the years, Magnetic Resonance Imaging (MRI) has been increasingly used to gain insights on the structural and functional abnormalities inherent to the disorder. Recent progress in machine learning together with the availability of large datasets now pave the way to capture complex relationships to make inferences at an individual level in the perspective of computer-aided diagnosis/prognosis or biomarkers discovery. Given the limitations of state-of-the-art sparse algorithms to produce stable and interpretable predictive signatures, we have pushed forward the regularization approaches extending classical algorithms with structural constraints issued from the known biological structure (spatial structure of the brain) in order to force the solution to adhere to biological priors, producing more plausible interpretable solutions. Such structured sparsity constraints have been leveraged to identify first, a neuroanatomical signature of schizophrenia and second a neuroimaging functional signature of hallucinations in patients with schizophrenia. Additionally, we also extended the popular PCA (Principal Component Analysis) with spatial regularization to identify interpretable patterns of the neuroimaging variability in either functional or anatomical meshes of the cortical surface
Vo, Xuan Thanh. "Apprentissage avec la parcimonie et sur des données incertaines par la programmation DC et DCA." Thesis, Université de Lorraine, 2015. http://www.theses.fr/2015LORR0193/document.
Повний текст джерелаIn this thesis, we focus on developing optimization approaches for solving some classes of optimization problems in sparsity and robust optimization for data uncertainty. Our methods are based on DC (Difference of Convex functions) programming and DCA (DC Algorithms) which are well-known as powerful tools in optimization. This thesis is composed of two parts: the first part concerns with sparsity while the second part deals with uncertainty. In the first part, a unified DC approximation approach to optimization problem involving the zero-norm in objective is thoroughly studied on both theoretical and computational aspects. We consider a common DC approximation of zero-norm that includes all standard sparse inducing penalty functions, and develop general DCA schemes that cover all standard algorithms in the field. Next, the thesis turns to the nonnegative matrix factorization (NMF) problem. We investigate the structure of the considered problem and provide appropriate DCA based algorithms. To enhance the performance of NMF, the sparse NMF formulations are proposed. Continuing this topic, we study the dictionary learning problem where sparse representation plays a crucial role. In the second part, we exploit robust optimization technique to deal with data uncertainty for two important problems in machine learning: feature selection in linear Support Vector Machines and clustering. In this context, individual data point is uncertain but varies in a bounded uncertainty set. Different models (box/spherical/ellipsoidal) related to uncertain data are studied. DCA based algorithms are developed to solve the robust problems
Vo, Xuan Thanh. "Apprentissage avec la parcimonie et sur des données incertaines par la programmation DC et DCA." Electronic Thesis or Diss., Université de Lorraine, 2015. http://www.theses.fr/2015LORR0193.
Повний текст джерелаIn this thesis, we focus on developing optimization approaches for solving some classes of optimization problems in sparsity and robust optimization for data uncertainty. Our methods are based on DC (Difference of Convex functions) programming and DCA (DC Algorithms) which are well-known as powerful tools in optimization. This thesis is composed of two parts: the first part concerns with sparsity while the second part deals with uncertainty. In the first part, a unified DC approximation approach to optimization problem involving the zero-norm in objective is thoroughly studied on both theoretical and computational aspects. We consider a common DC approximation of zero-norm that includes all standard sparse inducing penalty functions, and develop general DCA schemes that cover all standard algorithms in the field. Next, the thesis turns to the nonnegative matrix factorization (NMF) problem. We investigate the structure of the considered problem and provide appropriate DCA based algorithms. To enhance the performance of NMF, the sparse NMF formulations are proposed. Continuing this topic, we study the dictionary learning problem where sparse representation plays a crucial role. In the second part, we exploit robust optimization technique to deal with data uncertainty for two important problems in machine learning: feature selection in linear Support Vector Machines and clustering. In this context, individual data point is uncertain but varies in a bounded uncertainty set. Different models (box/spherical/ellipsoidal) related to uncertain data are studied. DCA based algorithms are developed to solve the robust problems
Zeng, Tieyong. "Études de Modèles Variationnels et Apprentissage de Dictionnaires." Phd thesis, Université Paris-Nord - Paris XIII, 2007. http://tel.archives-ouvertes.fr/tel-00178024.
Повний текст джерелаPilastre, Barbara. "Estimation parcimonieuse et apprentissage de dictionnaires pour la détection d'anomalies multivariées dans des données mixtes de télémesure satellites." Thesis, Toulouse, INPT, 2020. http://www.theses.fr/2020INPT0074.
Повний текст джерелаSpacecraft health monitoring and failure prevention are major issues in many fields and space industry has not escaped to this trend. Indeed, the proper conduct of satellite missions involves ensuring satellites good health and detect failures as soon as possible. This important task is performed by analyzing housekeeping telemetry data using anomaly detection methods. Housekeeping telemetry consist of sensors data recorded on board and received as time series describing the time evolution of various parameters. Each parameter is associated with physical quantity such as a temperature, a voltage or a pressure, or an equipement status. As conventional monitoring methods reach their limits, statistical machine learning methods have been studied to improve satellite telemetry monitoring via a semi-supervised learning: telemetry associated with normal operations of the spacecraft is learned to build a reference model. Then, more recent data is compared to this model in order to detect any potential anomalies. Most of the methods recently proposed focus on univariate anomaly detection for continuous parameters and handle telemetry parameters independently remove. The purpose of this thesis is to propose algorithms for multivariate anomaly detection which can handle mixed telemetry parameters jointly and take into account the correlations and relationships that may exist between them in order to detect univariate and multivariate anomalies. In this work we assume that telemetry signals can be approximated using few telemetry signals associated with normal satellite operations. This first hypothesis of sparsity justifies the use of sparse representation methods that will be studied throughout this thesis. This choice is also motivated by a second form of sparsity which is specific to satellite anomalies and reflect the fact that anomalies are rare and affect few parameters at the same time. In a first time, a multivariate anomaly detection algorithm based on a sparse estimation model is proposed. A weighted extension of the method which integrates external information is presented as well as a hyperparameter estimation method that has been developed to facilitate the operationnal use of the algorithm. In a second step, a sparse estimation model with a convolutional dictionary is proposed. The objective of this second method is to exploit the shiftinvariance property of convolutional dictionnaries and improve the detection. The proposed methods are finally evaluated on industrial use cases associated with real telemetry data and are compared to state-of-the-art approches
Ehsandoust, Bahram. "Séparation de Sources Dans des Mélanges non-Lineaires." Thesis, Université Grenoble Alpes (ComUE), 2018. http://www.theses.fr/2018GREAT033/document.
Повний текст джерелаBlind Source Separation (BSS) is a technique for estimating individual source components from their mixtures at multiple sensors, where the mixing model is unknown. Although it has been mathematically shown that for linear mixtures, under mild conditions, mutually independent sources can be reconstructed up to accepted ambiguities, there is not such theoretical basis for general nonlinear models. This is why there are relatively few results in the literature in this regard in the recent decades, which are focused on specific structured nonlinearities.In the present study, the problem is tackled using a novel approach utilizing temporal information of the signals. The original idea followed in this purpose is to study a linear time-varying source separation problem deduced from the initial nonlinear problem by derivations. It is shown that already-proposed counter-examples showing inefficiency of Independent Component Analysis (ICA) for nonlinear mixtures, loose their validity, considering independence in the sense of stochastic processes instead of simple random variables. Based on this approach, both nice theoretical results and algorithmic developments are provided. Even though these achievements are not claimed to be a mathematical proof for the separability of nonlinear mixtures, it is shown that given a few assumptions, which are satisfied in most practical applications, they are separable.Moreover, nonlinear BSS for two useful sets of source signals is also addressed: (1) spatially sparse sources and (2) Gaussian processes. Distinct BSS methods are proposed for these two cases, each of which has been widely studied in the literature and has been shown to be quite beneficial in modeling many practical applications.Concerning Gaussian processes, it is demonstrated that not all nonlinear mappings can preserve Gaussianity of the input. For example being restricted to polynomial functions, the only Gaussianity-preserving function is linear. This idea is utilized for proposing a linearizing algorithm which, cascaded by a conventional linear BSS method, separates polynomial mixturesof Gaussian processes.Concerning spatially sparse sources, it is shown that spatially sparsesources make manifolds in the observations space, and can be separated once the manifolds are clustered and learned. For this purpose, multiple manifold learning problem has been generally studied, whose results are not limited to the proposed BSS framework and can be employed in other topics requiring a similar issue
Mailhé, Boris. "Modèles et algorithmes pour la modélisation parcimonieuse de signaux de grande dimension." Phd thesis, Rennes 1, 2009. https://theses.hal.science/tel-00512559/fr/.
Повний текст джерелаThis thesis provides fast algorithms for sparse representations. Sparse representations consist in modelling the signal as a linear combination of a few atoms chosen among a redundant (more atoms than the signal dimension) dictionary. How to decompose a given signal over a given dictionary? This problem is NP-Complete. Existing suboptimal algorithms are either to slow to be applied on large signals or compute coarse approximations. We propose a new algorithm, LocOMP, that is both scalable and achieves good approximation quality. LocOMP only works with local dictionaries: the support of an atom is much shorter than the signal length. How to learn a dictionary on which a given class of signals can be decomposed? This problem is even more difficult: its resolution usually involves several sparse decompositions. We propose to improve the Olshausen-Field algorithm. It optimizes the dictionary via fixed step gradient descent. We show how to compute the optimal step. This makes the algorithm converge faster towards a better dictionary. These algorithms were applied to the study of atrial fibrillation. Atrial fibrillation is a common heart arrhythmia: the atria start vibrating instead of beating. One would like to observe I in the patient's ECG but the ECG is a mixture of fibrillation and ventricular activity. Our separation method is based on the learning of one dictionary for the fibrillation and one for the ventricular activity, both of them learnt on the patient's ECG
Mailhé, Boris. "Modèles et algorithmes pour la modélisation parcimonieuse de signaux de grande dimension." Phd thesis, Université Rennes 1, 2009. http://tel.archives-ouvertes.fr/tel-00512559.
Повний текст джерелаAghaei, Mazaheri Jérémy. "Représentations parcimonieuses et apprentissage de dictionnaires pour la compression et la classification d'images satellites." Thesis, Rennes 1, 2015. http://www.theses.fr/2015REN1S028/document.
Повний текст джерелаThis thesis explores sparse representation and dictionary learning methods to compress and classify satellite images. Sparse representations consist in approximating a signal by a linear combination of a few columns, known as atoms, from a dictionary, and thus representing it by only a few non-zero coefficients contained in a sparse vector. In order to improve the quality of the representations and to increase their sparsity, it is interesting to learn the dictionary. The first part of the thesis presents a state of the art about sparse representations and dictionary learning methods. Several applications of these methods are explored. Some image compression standards are also presented. The second part deals with the learning of dictionaries structured in several levels, from a tree structure to an adaptive structure, and their application to the compression of satellite images, by integrating them in an adapted coding scheme. Finally, the third part is about the use of learned structured dictionaries for the classification of satellite images. A method to estimate the Modulation Transfer Function (MTF) of the instrument used to capture an image is studied. A supervised classification algorithm, using structured dictionaries made discriminant between classes during the learning, is then presented in the scope of scene recognition in a picture
Lounici, Karim. "Estimation Statistique En Grande Dimension, Parcimonie et Inégalités D'Oracle." Phd thesis, Université Paris-Diderot - Paris VII, 2009. http://tel.archives-ouvertes.fr/tel-00435917.
Повний текст джерелаMountassir, Mahjoub El. "Surveillance d'intégrité des structures par apprentissage statistique : application aux structures tubulaires." Electronic Thesis or Diss., Université de Lorraine, 2019. http://docnum.univ-lorraine.fr/ulprive/DDOC_T_2019_0047_EL_MOUNTASSIR.pdf.
Повний текст джерелаTo ensure better working conditions of civil and engineering structures, inspections must be made on a regular basis. However, these inspections could be labor-intensive and cost-consuming. In this context, structural health monitoring (SHM) systems using permanently attached transducers were proposed to ensure continuous damage diagnostic of these structures. In SHM, damage detection is generally based on comparison between the healthy state signals and the current signals. Nevertheless, the environmental and operational conditions will have an effect on the healthy state signals. If these effects are not taken into account they would result in false indication of damage (false alarm). In this thesis, classical machine learning methods used for damage detection have been applied in the case of pipelines. The effects of some measurements parameters on the robustness of these methods have been investigated. Afterthat, two approaches were proposed for damage diagnostic depending on the database of reference signals. If this database contains large variation of these EOCs, a sparse estimation of the current signal is calculated. Then, the estimation error is used as an indication of the presence of damage. Otherwise, if this database is acquired at limited range of EOCs, moving window PCA can be applied to update the model of the healthy state provided that the EOCs show slow and continuous variation. In both approaches, damage localization was ensured using a sliding window over the damaged pipe signal
Mountassir, Mahjoub El. "Surveillance d'intégrité des structures par apprentissage statistique : application aux structures tubulaires." Thesis, Université de Lorraine, 2019. http://www.theses.fr/2019LORR0047.
Повний текст джерелаTo ensure better working conditions of civil and engineering structures, inspections must be made on a regular basis. However, these inspections could be labor-intensive and cost-consuming. In this context, structural health monitoring (SHM) systems using permanently attached transducers were proposed to ensure continuous damage diagnostic of these structures. In SHM, damage detection is generally based on comparison between the healthy state signals and the current signals. Nevertheless, the environmental and operational conditions will have an effect on the healthy state signals. If these effects are not taken into account they would result in false indication of damage (false alarm). In this thesis, classical machine learning methods used for damage detection have been applied in the case of pipelines. The effects of some measurements parameters on the robustness of these methods have been investigated. Afterthat, two approaches were proposed for damage diagnostic depending on the database of reference signals. If this database contains large variation of these EOCs, a sparse estimation of the current signal is calculated. Then, the estimation error is used as an indication of the presence of damage. Otherwise, if this database is acquired at limited range of EOCs, moving window PCA can be applied to update the model of the healthy state provided that the EOCs show slow and continuous variation. In both approaches, damage localization was ensured using a sliding window over the damaged pipe signal
Varasteh, Yazdi Saeed. "Représentations parcimonieuses et apprentissage de dictionnaires pour la classification et le clustering de séries temporelles." Thesis, Université Grenoble Alpes (ComUE), 2018. http://www.theses.fr/2018GREAM062/document.
Повний текст джерелаLearning dictionary for sparse representing time series is an important issue to extract latent temporal features, reveal salient primitives and sparsely represent complex temporal data. This thesis addresses the sparse coding and dictionary learning problem for time series classification and clustering under time warp. For that, we propose a time warp invariant sparse coding and dictionary learning framework where both input samples and atoms define time series of different lengths that involve varying delays.In the first part, we formalize an L0 sparse coding problem and propose a time warp invariant orthogonal matching pursuit based on a new cosine maximization time warp operator. For the dictionary learning stage, a non linear time warp invariant kSVD (TWI-kSVD) is proposed. Thanks to a rotation transformation between each atom and its sibling atoms, a singular value decomposition is used to jointly approximate the coefficients and update the dictionary, similar to the standard kSVD. In the second part, a time warp invariant dictionary learning for time series clustering is formalized and a gradient descent solution is proposed.The proposed methods are confronted to major shift invariant, convolved and kernel dictionary learning methods on several public and real temporal data. The conducted experiments show the potential of the proposed frameworks to efficiently sparse represent, classify and cluster time series under time warp
Kasper, Kévin. "Apprentissage d'estimateurs sans modèle avec peu de mesures - Application à la mécanique des fluides." Thesis, Université Paris-Saclay (ComUE), 2016. http://www.theses.fr/2016SACLN029/document.
Повний текст джерелаThis thesis deals with sparsity promoting techniques in order to produce efficient estimators relying only on a small amount of measurements given by sensors. These sensor locations are crucial to the estimators and have to be chosen meticulously. The proposed methods do not require dynamical models and are instead based on a collection of snapshots of the field of interest. This learning sequence can be acquired through measurements on the real system or through numerical simulation. By relying only on a learning sequence, and not on dynamical models, the proposed methods become general and applicable to a variety of systems.These techniques are illustrated on the 2-D fluid flow around a cylindrical body. The pressure field in the neighbourhood of the cylinder has to be estimated from a limited amount of surface pressure measurements. For a given arrangement of the sensors, efficient estimators suited to these locations are proposed. These estimators fully harness the information given by the limited amount of sensors by manipulating sparse representations and classes. Cases where the measurements are no longer made on the field to be estimated can also be considered. A sensor placement algorithm is proposed in order to improve the performances of the estimators.Multiple extensions are discussed : incorporating past measurements, past control inputs, recovering a field non-linearly related to the measurements, estimating a vectorial field, etc
Chan, wai tim Stefen. "Apprentissage supervisé d’une représentation multi-couches à base de dictionnaires pour la classification d’images et de vidéos." Thesis, Université Grenoble Alpes (ComUE), 2016. http://www.theses.fr/2016GREAT089/document.
Повний текст джерелаIn the recent years, numerous works have been published on dictionary learning and sparse coding. They were initially used in image reconstruction and image restoration tasks. Recently, researches were interested in the use of dictionaries for classification tasks because of their capability to represent underlying patterns in images. Good results have been obtained in specific conditions: centered objects of interest, homogeneous sizes and points of view.However, without these constraints, the performances are dropping.In this thesis, we are interested in finding good dictionaries for classification.The learning methods classically used for dictionaries rely on unsupervised learning. Here, we are going to study how to perform supervised dictionary learning.In order to push the performances further, we introduce a multilayer architecture for dictionaries. The proposed architecture is based on the local description of an input image and its transformation thanks to a succession of encoding and processing steps. It outputs a vector of features effective for classification.The learning method we developed is based on the backpropagation algorithm which allows a joint learning of the different dictionaries and an optimization solely with respect to the classification cost.The proposed architecture has been tested on MNIST, CIFAR-10 and STL-10 datasets with good results compared to other dicitonary-based methods. The proposed architecture can be extended to video analysis
Tran, Khanh-Hung. "Semi-supervised dictionary learning and Semi-supervised deep neural network." Thesis, université Paris-Saclay, 2021. http://www.theses.fr/2021UPASP014.
Повний текст джерелаSince the 2010's, machine learning (ML) has been one of the topics that attract a lot of attention from scientific researchers. Many ML models have been demonstrated their ability to produce excellent results in various fields such as Computer Vision, Natural Language Processing, Robotics... However, most of these models use supervised learning, which requires a massive annotation. Therefore, the objective of this thesis is to study and to propose semi-supervised learning approaches that have many advantages over supervised learning. Instead of directly applying a semi-supervised classifier on the original representation of data, we rather use models that integrate a representation learning stage before the classification stage, to better adapt to the non-linearity of the data. In the first step, we revisit tools that allow us to build our semi-supervised models. First, we present two types of model that possess representation learning in their architecture: dictionary learning and neural network, as well as the optimization methods for each type of model. Moreover, in the case of neural network, we specify the problem with adversarial examples. Then, we present the techniques that often accompany with semi-supervised learning such as variety learning and pseudo-labeling. In the second part, we work on dictionary learning. We synthesize generally three steps to build a semi-supervised model from a supervised model. Then, we propose our semi-supervised model to deal with the classification problem typically in the case of a low number of training samples (including both labelled and non-labelled samples). On the one hand, we apply the preservation of the data structure from the original space to the sparse code space (manifold learning), which is considered as regularization for sparse codes. On the other hand, we integrate a semi-supervised classifier in the sparse code space. In addition, we perform sparse coding for test samples by taking into account also the preservation of the data structure. This method provides an improvement on the accuracy rate compared to other existing methods. In the third step, we work on neural network models. We propose an approach called "manifold attack" which allows reinforcing manifold learning. This approach is inspired from adversarial learning : finding virtual points that disrupt the cost function on manifold learning (by maximizing it) while fixing the model parameters; then the model parameters are updated by minimizing this cost function while fixing these virtual points. We also provide criteria for limiting the space to which the virtual points belong and the method for initializing them. This approach provides not only an improvement on the accuracy rate but also a significant robustness to adversarial examples. Finally, we analyze the similarities and differences, as well as the advantages and disadvantages between dictionary learning and neural network models. We propose some perspectives on both two types of models. In the case of semi-supervised dictionary learning, we propose some techniques inspired by the neural network models. As for the neural network, we propose to integrate manifold attack on generative models
Barthelemy, Quentin. "Représentations parcimonieuses pour les signaux multivariés." Thesis, Grenoble, 2013. http://www.theses.fr/2013GRENU008/document.
Повний текст джерелаIn this thesis, we study approximation and learning methods which provide sparse representations. These methods allow to analyze very redundant data-bases thanks to learned atoms dictionaries. Being adapted to studied data, they are more efficient in representation quality than classical dictionaries with atoms defined analytically. We consider more particularly multivariate signals coming from the simultaneous acquisition of several quantities, as EEG signals or 2D and 3D motion signals. We extend sparse representation methods to the multivariate model, to take into account interactions between the different components acquired simultaneously. This model is more flexible that the common multichannel one which imposes a hypothesis of rank 1. We study models of invariant representations: invariance to temporal shift, invariance to rotation, etc. Adding supplementary degrees of freedom, each kernel is potentially replicated in an atoms family, translated at all samples, rotated at all orientations, etc. So, a dictionary of invariant kernels generates a very redundant atoms dictionary, thus ideal to represent the redundant studied data. All these invariances require methods adapted to these models. Temporal shift-invariance is an essential property for the study of temporal signals having a natural temporal variability. In the 2D and 3D rotation invariant case, we observe the efficiency of the non-oriented approach over the oriented one, even when data are not revolved. Indeed, the non-oriented model allows to detect data invariants and assures the robustness to rotation when data are revolved. We also observe the reproducibility of the sparse decompositions on a learned dictionary. This generative property is due to the fact that dictionary learning is a generalization of K-means. Moreover, our representations have many invariances that is ideal to make classification. We thus study how to perform a classification adapted to the shift-invariant model, using shift-consistent pooling functions
Isaac, Yoann. "Représentations redondantes pour les signaux d’électroencéphalographie." Thesis, Paris 11, 2015. http://www.theses.fr/2015PA112072/document.
Повний текст джерелаThe electroencephalography measures the brain activity by recording variations of the electric field on the surface of the skull. This measurement is usefull in various applications like medical diagnosis, analysis of brain functionning or whithin brain-computer interfaces. Numerous studies have tried to develop methods for analyzing these signals in order to extract various components of interest, however, none of them allows to extract them with sufficient reliabilty. This thesis focuses on the development of approaches considering redundant (overcomoplete) representations for these signals. During the last years, these representations have been shown particularly efficient to describe various classes of signals due to their flexibility. Obtaining such representations for EEG presents some difficuties due to the low signal-to-noise ratio of these signals. We propose in this study to overcome them by guiding the methods considered to physiologically plausible representations thanks to well-suited regularizations. These regularizations are built from prior knowledge about the spatial and temporal properties of these signals. For each regularization, an algorithm is proposed to solve the optimization problem allowing to obtain the targeted representations. The evaluation of the proposed EEG signals approaches highlights their effectiveness in representing them
Raja, Suleiman Raja Fazliza. "Méthodes de detection robustes avec apprentissage de dictionnaires. Applications à des données hyperspectrales." Thesis, Nice, 2014. http://www.theses.fr/2014NICE4121/document.
Повний текст джерелаThis Ph.D dissertation deals with a "one among many" detection problem, where one has to discriminate between pure noise under H0 and one among L known alternatives under H1. This work focuses on the study and implementation of robust reduced dimension detection tests using optimized dictionaries. These detection methods are associated with the Generalized Likelihood Ratio test. The proposed approaches are principally assessed on hyperspectral data. In the first part, several technical topics associated to the framework of this dissertation are presented. The second part highlights the theoretical and algorithmic aspects of the proposed methods. Two issues linked to the large number of alternatives arise in this framework. In this context, we propose dictionary learning techniques based on a robust criterion that seeks to minimize the maximum power loss (type minimax). In the case where the learned dictionary has K = 1 column, we show that the exact solution can be obtained. Then, we propose in the case K > 1 three minimax learning algorithms. Finally, the third part of this manuscript presents several applications. The principal application regards astrophysical hyperspectral data of the Multi Unit Spectroscopic Explorer instrument. Numerical results show that the proposed algorithms are robust and in the case K > 1 they allow to increase the minimax detection performances over the K = 1 case. Other possible applications such as worst-case recognition of faces and handwritten digits are presented
Gerchinovitz, Sébastien. "Prédiction de suites individuelles et cadre statistique classique : étude de quelques liens autour de la régression parcimonieuse et des techniques d'agrégation." Phd thesis, Université Paris Sud - Paris XI, 2011. http://tel.archives-ouvertes.fr/tel-00653550.
Повний текст джерелаLe, Folgoc Loïc. "Apprentissage statistique pour la personnalisation de modèles cardiaques à partir de données d’imagerie." Thesis, Nice, 2015. http://www.theses.fr/2015NICE4098/document.
Повний текст джерелаThis thesis focuses on the calibration of an electromechanical model of the heart from patient-specific, image-based data; and on the related task of extracting the cardiac motion from 4D images. Long-term perspectives for personalized computer simulation of the cardiac function include aid to the diagnosis, aid to the planning of therapy and prevention of risks. To this end, we explore tools and possibilities offered by statistical learning. To personalize cardiac mechanics, we introduce an efficient framework coupling machine learning and an original statistical representation of shape & motion based on 3D+t currents. The method relies on a reduced mapping between the space of mechanical parameters and the space of cardiac motion. The second focus of the thesis is on cardiac motion tracking, a key processing step in the calibration pipeline, with an emphasis on quantification of uncertainty. We develop a generic sparse Bayesian model of image registration with three main contributions: an extended image similarity term, the automated tuning of registration parameters and uncertainty quantification. We propose an approximate inference scheme that is tractable on 4D clinical data. Finally, we wish to evaluate the quality of uncertainty estimates returned by the approximate inference scheme. We compare the predictions of the approximate scheme with those of an inference scheme developed on the grounds of reversible jump MCMC. We provide more insight into the theoretical properties of the sparse structured Bayesian model and into the empirical behaviour of both inference schemes
Le, Van Luong. "Identification de systèmes dynamiques hybrides : géométrie, parcimonie et non-linéarités." Electronic Thesis or Diss., Université de Lorraine, 2013. http://www.theses.fr/2013LORR0102.
Повний текст джерелаIn automatic control, obtaining a model is always the cornerstone of the synthesis procedures such as controller design, fault detection or prediction... This thesis deals with the identification of a class of complex systems, hybrid dynamical systems. These systems involve the interaction of continuous and discrete behaviors. The goal is to build a model from experimental measurements of the system inputs and outputs. A new approach for the identification of linear hybrid systems based on the geometric properties of hybrid systems in the parameter space is proposed. A new algorithm is then proposed to recover the sparsest solutions of underdetermined systems of linear equations. This allows us to improve an identification approach based on the error sparsification. In addition, new approaches based on kernel models are proposed for the identification of nonlinear hybrid systems and piecewise smooth systems
Moscu, Mircea. "Inférence distribuée de topologie de graphe à partir de flots de données." Thesis, Université Côte d'Azur, 2020. http://www.theses.fr/2020COAZ4081.
Повний текст джерелаThe second decade of the current millennium can be summarized in one short phrase: the advent of data. There has been a surge in the number of data sources: from audio-video streaming, social networks and the Internet of Things, to smartwatches, industrial equipment and personal vehicles, just to name a few. More often than not, these sources form networks in order to exchange information. As a direct consequence, the field of Graph Signal Processing has been thriving and evolving. Its aim: process and make sense of all the surrounding data deluge.In this context, the main goal of this thesis is developing methods and algorithms capable of using data streams, in a distributed fashion, in order to infer the underlying networks that link these streams. Then, these estimated network topologies can be used with tools developed for Graph Signal Processing in order to process and analyze data supported by graphs. After a brief introduction followed by motivating examples, we first develop and propose an online, distributed and adaptive algorithm for graph topology inference for data streams which are linearly dependent. An analysis of the method ensues, in order to establish relations between performance and the input parameters of the algorithm. We then run a set of experiments in order to validate the analysis, as well as compare its performance with that of another proposed method of the literature.The next contribution is in the shape of an algorithm endowed with the same online, distributed and adaptive capacities, but adapted to inferring links between data that interact non-linearly. As such, we propose a simple yet effective additive model which makes use of the reproducing kernel machinery in order to model said nonlinearities. The results if its analysis are convincing, while experiments ran on biomedical data yield estimated networks which exhibit behavior predicted by medical literature.Finally, a third algorithm proposition is made, which aims to improve the nonlinear model by allowing it to escape the constraints induced by additivity. As such, the newly proposed model is as general as possible, and makes use of a natural and intuitive manner of imposing link sparsity, based on the concept of partial derivatives. We analyze this proposed algorithm as well, in order to establish stability conditions and relations between its parameters and its performance. A set of experiments are ran, showcasing how the general model is able to better capture nonlinear links in the data, while the estimated networks behave coherently with previous estimates
Nasser, Khalafallah Mahmoud Lamees. "A dictionary-based denoising method toward a robust segmentation of noisy and densely packed nuclei in 3D biological microscopy images." Electronic Thesis or Diss., Sorbonne université, 2019. https://accesdistant.sorbonne-universite.fr/login?url=https://theses-intra.sorbonne-universite.fr/2019SORUS283.pdf.
Повний текст джерелаCells are the basic building blocks of all living organisms. All living organisms share life processes such as growth and development, movement, nutrition, excretion, reproduction, respiration and response to the environment. In cell biology research, understanding cells structure and function is essential for developing and testing new drugs. In addition, cell biology research provides a powerful tool to study embryo development. Furthermore, it helps the scientific research community to understand the effects of mutations and various diseases. Time-Lapse Fluorescence Microscopy (TLFM) is one of the most appreciated imaging techniques which can be used in live-cell imaging experiments to quantify various characteristics of cellular processes, i.e., cell survival, proliferation, migration, and differentiation. In TLFM imaging, not only spatial information is acquired, but also temporal information obtained by repeating imaging of a labeled sample at specific time points, as well as spectral information, that produces up to five-dimensional (X, Y, Z + Time + Channel) images. Typically, the generated datasets consist of several (hundreds or thousands) images, each containing hundreds to thousands of objects to be analyzed. To perform high-throughput quantification of cellular processes, nuclei segmentation and tracking should be performed in an automated manner. Nevertheless, nuclei segmentation and tracking are challenging tasks due to embedded noise, intensity inhomogeneity, shape variation as well as a weak boundary of nuclei. Although several nuclei segmentation approaches have been reported in the literature, dealing with embedded noise remains the most challenging part of any segmentation algorithm. We propose a novel 3D denoising algorithm, based on unsupervised dictionary learning and sparse representation, that can both enhance very faint and noisy nuclei, in addition, it simultaneously detects nuclei position accurately. Furthermore, our method is based on a limited number of parameters, with only one being critical, which is the approximate size of the objects of interest. The framework of the proposed method comprises image denoising, nuclei detection, and segmentation. In the denoising step, an initial dictionary is constructed by selecting random patches from the raw image then an iterative technique is implemented to update the dictionary and obtain the final one which is less noisy. Next, a detection map, based on the dictionary coefficients used to denoise the image, is used to detect marker points. Afterward, a thresholding-based approach is proposed to get the segmentation mask. Finally, a marker-controlled watershed approach is used to get the final nuclei segmentation result. We generate 3D synthetic images to study the effect of the few parameters of our method on cell nuclei detection and segmentation, and to understand the overall mechanism for selecting and tuning the significant parameters of the several datasets. These synthetic images have low contrast and low signal to noise ratio. Furthermore, they include touching spheres where these conditions simulate the same characteristics exist in the real datasets. The proposed framework shows that integrating our denoising method along with classical segmentation method works properly in the context of the most challenging cases. To evaluate the performance of the proposed method, two datasets from the cell tracking challenge are extensively tested. Across all datasets, the proposed method achieved very promising results with 96.96% recall for the C.elegans dataset. Besides, in the Drosophila dataset, our method achieved very high recall (99.3%)
Phan, Duy Nhat. "Algorithmes basés sur la programmation DC et DCA pour l’apprentissage avec la parcimonie et l’apprentissage stochastique en grande dimension." Electronic Thesis or Diss., Université de Lorraine, 2016. http://www.theses.fr/2016LORR0235.
Повний текст джерелаThese days with the increasing abundance of data with high dimensionality, high dimensional classification problems have been highlighted as a challenge in machine learning community and have attracted a great deal of attention from researchers in the field. In recent years, sparse and stochastic learning techniques have been proven to be useful for this kind of problem. In this thesis, we focus on developing optimization approaches for solving some classes of optimization problems in these two topics. Our methods are based on DC (Difference of Convex functions) programming and DCA (DC Algorithms) which are wellknown as one of the most powerful tools in optimization. The thesis is composed of three parts. The first part tackles the issue of variable selection. The second part studies the problem of group variable selection. The final part of the thesis concerns the stochastic learning. In the first part, we start with the variable selection in the Fisher's discriminant problem (Chapter 2) and the optimal scoring problem (Chapter 3), which are two different approaches for the supervised classification in the high dimensional setting, in which the number of features is much larger than the number of observations. Continuing this study, we study the structure of the sparse covariance matrix estimation problem and propose four appropriate DCA based algorithms (Chapter 4). Two applications in finance and classification are conducted to illustrate the efficiency of our methods. The second part studies the L_p,0regularization for the group variable selection (Chapter 5). Using a DC approximation of the L_p,0norm, we indicate that the approximate problem is equivalent to the original problem with suitable parameters. Considering two equivalent reformulations of the approximate problem we develop DCA based algorithms to solve them. Regarding applications, we implement the proposed algorithms for group feature selection in optimal scoring problem and estimation problem of multiple covariance matrices. In the third part of the thesis, we introduce a stochastic DCA for large scale parameter estimation problems (Chapter 6) in which the objective function is a large sum of nonconvex components. As an application, we propose a special stochastic DCA for the loglinear model incorporating latent variables
Barthélemy, Quentin. "Représentations parcimonieuses pour les signaux multivariés." Phd thesis, Université de Grenoble, 2013. http://tel.archives-ouvertes.fr/tel-00853362.
Повний текст джерелаDantas, Cássio Fraga. "Accelerating sparse inverse problems using structured approximations." Thesis, Rennes 1, 2019. http://www.theses.fr/2019REN1S065.
Повний текст джерелаAs the quantity and size of available data grow, the existing algorithms for solving sparse inverse problems can become computationally intractable. In this work, we explore two main strategies for accelerating such algorithms. First, we study the use of structured dictionaries which are fast to operate with. A particular family of dictionaries, written as a sum of Kronecker products, is proposed. Then, we develop stable screening tests, which can safely identify and discard useless atoms (columns of the dictionary matrix which do not correspond to the solution support), despite manipulating approximate dictionaries
Le, Van Luong. "Identification de systèmes dynamiques hybrides : géométrie, parcimonie et non-linéarités." Phd thesis, Université de Lorraine, 2013. http://tel.archives-ouvertes.fr/tel-00874283.
Повний текст джерела