Índice
Literatura académica sobre el tema "Régularisation implicite"
Crea una cita precisa en los estilos APA, MLA, Chicago, Harvard y otros
Consulte las listas temáticas de artículos, libros, tesis, actas de conferencias y otras fuentes académicas sobre el tema "Régularisation implicite".
Junto a cada fuente en la lista de referencias hay un botón "Agregar a la bibliografía". Pulsa este botón, y generaremos automáticamente la referencia bibliográfica para la obra elegida en el estilo de cita que necesites: APA, MLA, Harvard, Vancouver, Chicago, etc.
También puede descargar el texto completo de la publicación académica en formato pdf y leer en línea su resumen siempre que esté disponible en los metadatos.
Artículos de revistas sobre el tema "Régularisation implicite"
Neveu, Emilie, Laurent Debreu y François-Xavier Le Dimet. "Multigrid methods and data assimilation ― Convergence study and first experiments on non-linear equations". Revue Africaine de la Recherche en Informatique et Mathématiques Appliquées Volume 14 - 2011 - Special... (21 de agosto de 2011). http://dx.doi.org/10.46298/arima.1944.
Texto completoTesis sobre el tema "Régularisation implicite"
Belhachmi, Ayoub. "Une méthode implicite pour la construction des modèles géologiques complexes via une interpolation à l'aide des splines et une régularisation basée sur les équations aux dérivées partielles". Electronic Thesis or Diss., Université Côte d'Azur, 2024. http://www.theses.fr/2024COAZ5000.
Texto completoThe construction of a geological numerical model is a key step in the study and exploration of the subsurface. These models are constructed from seismic or well data, which consist of data points associated with values corresponding to their geological ages. This task involves constructing an implicit function, known also as stratigraphic function, which interpolates this set of data points. Often the available data are sparse and noisy, which makes this task difficult, mainly for reservoirs where the geological structures are complex with distinct discontinuities and unconformities. To address this, the interpolation problem is typically supplemented with a regularization term that enforces a regular behaviour of the implicit function. In this thesis, we propose a new method to compute the stratigraphic function that represents geological layers in arbitrary settings. In this method, the data are interpolated by piecewise quadratic C^1 Powell-Sabin splines and the function can be regularized via many regularization energies. The method is discretized in finite elements on a triangular mesh conforming to the geological faults. Compared to classical interpolation methods, the use of piecewise quadratic splines has two major advantages. First, a better handling of stratigraphic surfaces with strong curvatures. Second, a reduction in mesh resolution, while generating surfaces of higher smoothness and regularity.The regularization of the function is the most difficult component of any implicit modeling approach. Often, classical methods produce inconsistent geological models, in particular for data with high thickness variation, and bubble effects are generally observed. To handle this problem, we introduce two new regularization energies that are linked to two fundamental PDEs, in their general form with spatially varying coefficients. These PDEs are the anisotropic diffusion equation and the equation that describes the bending of an anisotropic thin plate. In the first approach, the diffusion tensor is introduced and iteratively adapted to the variations and anisotropy of the data. In the second, the rigidity tensor is iteratively adapted to the variations and anisotropy in the data. We demonstrate the effectiveness of the proposed methods in 2D, specifically on cross-sections of geological models with complex fault networks and thickness variations in the layers
Ayme, Alexis. "Supervised learning with missing data : a non-asymptotic point of view". Electronic Thesis or Diss., Sorbonne université, 2024. http://www.theses.fr/2024SORUS252.
Texto completoMissing values are common in most real-world data sets due to the combination of multiple sources andinherently missing information, such as sensor failures or unanswered survey questions. The presenceof missing values often prevents the application of standard learning algorithms. This thesis examinesmissing values in a prediction context, aiming to achieve accurate predictions despite the occurrence ofmissing data in both training and test datasets. The focus of this thesis is to theoretically analyze specific algorithms to obtain finite-sample guarantees. We derive minimax lower bounds on the excess risk of linear predictions in presence of missing values.Such lower bounds depend on the distribution of the missing pattern, and can grow exponentially withthe dimension. We propose a very simple method consisting in applying Least-Square procedure onthe most frequent missing patterns only. Such a simple method turns out to be near minimax-optimalprocedure, which departs from the Least-Square algorithm applied to all missing patterns. Followingthis, we explore the impute-then-regress method, where imputation is performed using the naive zeroimputation, and the regression step is carried out via linear models, whose parameters are learned viastochastic gradient descent. We demonstrate that this very simple method offers strong finite-sampleguarantees in high-dimensional settings. Specifically, we show that the bias of this method is lowerthan the bias of ridge regression. As ridge regression is often used in high dimensions, this proves thatthe bias of missing data (via zero imputation) is negligible in some high-dimensional settings. Thesefindings are illustrated using random features models, which help us to precisely understand the role ofdimensionality. Finally, we study different algorithm to handle linear classification in presence of missingdata (logistic regression, perceptron, LDA). We prove that LDA is the only model that can be valid forboth complete and missing data for some generic settings
Samozino, Marie. "Voronoi Centred Radial Basis Functions". Phd thesis, Université de Nice Sophia-Antipolis, 2007. http://tel.archives-ouvertes.fr/tel-00336379.
Texto completoLa surface est reconstruite comme le niveau zéro d'une fonction. Représenter une surface implicitement en utilisant des fonctions de base radiales (Radial Basis Functions) est devenu une approche standard ces dix dernières années. Une problématique intéressante est la réduction du nombre de fonctions de base pour obtenir une représentation la plus compacte possible et réduire les temps d'évaluation.
Réduire le nombre de fonctions de base revient à réduire le nombre de points (centres) sur lesquels elles sont centrées. L'objectif que l'on s'est fixé consiste à sélectionner un "petit" ensemble de centres, les plus pertinents possible. Pour réduire le nombre de centres tout en gardant un maximum d'information, nous nous sommes affranchis de la correspondance entre centres des fonctions et points de donnée, qui est imposée dans la quasi-totalité des approches RBF. Au contraire, nous avons décidé de placer les centres sur l'axe médian de l'ensemble des points de donnée et de montrer que ce choix était approprié.
Pour cela, nous avons utilisé les outils donnés par la géométrie algorithmique et approximé l'axe médian par un sous-ensemble des sommets du diagramme de Voronoi des points de donnée. Nous avons aussi proposé deux approches différentes qui échantillonnent de manière appropriée l'axe médian pour adapter le niveau de détail de la surface reconstruite au budget de centres alloué par l'utilisateur.
Chamoret, Dominique. "Modélisation du contact : nouvelles approches numériques". Ecully, Ecole centrale de Lyon, 2002. http://bibli.ec-lyon.fr/exl-doc/TH_T1897_dchamoret.pdf.
Texto completoThis work has been done in collaboration between the Laboratoire de Tribologie et Dynamique des Systèmes (UMR 5513 CNRS/ECL/ENISE) and the company ESI Sofware. The aim is to develop the modelling of 3D contact in an implicit approach and to implement algorithms likely to adapt to most of situations met in industry. With this intention we propose three new algorithms concerning the regularization of contact surfaces, the adaptation of the penalty parameter and the adjustment of the load step. The numerical treatment of contact problems generates many difficulties. These problems come from strong geometric and material nonlinearities. Using the finite element method, the contact interface is represented by a surface only piecewise differentiable. The strategy we develop determines a smooth contact surface by using only the data of the nodes of the initial finite element mesh thanks to the technique of diffuse approximation. We have chosen to use the penalty method to take into account contact constraints but the optimum choice of the penalty parameter is often very difficult because it depends on the local rigidity of the structure. We propose an original strategy to determine an optimal parameter based on the notion of acceptable interpenetration between bodies in contact. The last algorithm proposed relates to the adjustment of the load step. The main idea of the method we develop is to limit the number of changes of contact status in each load step
Samozino, Marie. "Voronoi Centered Radial Basis Functions". Phd thesis, Université de Nice Sophia-Antipolis, 2007. http://tel.archives-ouvertes.fr/tel-00178274.
Texto completoFoucault, Alexandre. "Modélisation du comportement cyclique des ouvrages en terre intégrant des techniques de régularisation". Phd thesis, Ecole Centrale Paris, 2010. http://tel.archives-ouvertes.fr/tel-00534665.
Texto completoLaporte, 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/.
Texto completoLearning-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
Estecahandy, Elodie. "Contribution à l'analyse mathématique et à la résolution numérique d'un problème inverse de scattering élasto-acoustique". Phd thesis, Université de Pau et des Pays de l'Adour, 2013. http://tel.archives-ouvertes.fr/tel-00880628.
Texto completo