Estimation and feature selection in high dimensional mixtures of experts models (2)
➤ Gửi thông báo lỗi ⚠️ Báo cáo tài liệu vi phạmNội dung chi tiết: Estimation and feature selection in high dimensional mixtures of experts models (2)
Estimation and feature selection in high dimensional mixtures of experts models (2)
Normandie UniversitéTHESEPour obtenir le diplôme de doctoratSpécialité MATHEMATIQUESPréparée au sein de 1'Université de Caen NormandieEstimation and F Estimation and feature selection in high dimensional mixtures of experts models (2) Feature Selection in High-Dimensional Mixtures-of-Experts ModelsPresentee et soutenue par Bao Tuyen HUYNHThese soutenue publiquement le 03/12/2019 devant le jury compose deMme FLORENCE FORBESDirecteurde recherche à I'INRIA, INRIA GrenobleRapporteur du juryM. JULIEN JACQUESProfesseur des universités. Estimation and feature selection in high dimensional mixtures of experts models (2) Université Lumière Lyon 2Rapporteur du juryMme ÊMILIE DEVIJVERCharge de recherche. Laboratoire LIG GrenobleMembre du juryM. MICHAEL FOPMaitre de coníEstimation and feature selection in high dimensional mixtures of experts models (2)
érences. University College DublinMembre du juryM. FAICEL CHAMROUKHIProfesseur des umversités. Université Caen NormandieDirecteur de theseThese dirigéNormandie UniversitéTHESEPour obtenir le diplôme de doctoratSpécialité MATHEMATIQUESPréparée au sein de 1'Université de Caen NormandieEstimation and F Estimation and feature selection in high dimensional mixtures of experts models (2) uld not have been achieved without the help and support of many friends, colleagues, teachers and my family. First. I would like to thank my advisor Fated Chamroukhi for his guidance with an interesting topic. I appreciate all of the time you have spent discussing to all of my ideas, and all of the Estimation and feature selection in high dimensional mixtures of experts models (2) effort that you have put into fixing my bad grammar. Most of all. I appreciate your friendship.A very special thanks to Mr. Eric Ricard for his kindneEstimation and feature selection in high dimensional mixtures of experts models (2)
ss and support as a director of LMNO Lab.I would like to thank Mme. Florence Forbes and Mr. Julien Jacques for reviewing my thesis. I am also gratefulNormandie UniversitéTHESEPour obtenir le diplôme de doctoratSpécialité MATHEMATIQUESPréparée au sein de 1'Université de Caen NormandieEstimation and F Estimation and feature selection in high dimensional mixtures of experts models (2) rnaud. Cesar. Guillaume, Etienne, Julien. Hung. Mostafa. Nacer. Pablo, Thanh. Thien and Tin. Many thanks to you, guys. It has been a pleasure to spend the last throe years or more getting to know you.Very special thanks to my teacher Dang Phuoc-Huy. Your enthusiasm for Statistics, your wealth of kno Estimation and feature selection in high dimensional mixtures of experts models (2) wledge, and your diligence inspires me to learn more ami work harder, in an attempt, to emulate your achievements. Without your help and support I canEstimation and feature selection in high dimensional mixtures of experts models (2)
not finalize this work. I also extend my thanks to anh Vu and CÔ Phuong for their kindness as my brother and sister.Finally, to my parents, my parentsNormandie UniversitéTHESEPour obtenir le diplôme de doctoratSpécialité MATHEMATIQUESPréparée au sein de 1'Université de Caen NormandieEstimation and F Estimation and feature selection in high dimensional mixtures of experts models (2) I am very fortunate to be part of our happy family. Last but not least, thank you my three little kids. You always make me feel happy and I love you with all my heart.Caen. September 30. 2019Huynh Bao TuycnAbstractThe stat istical analysis of heterogeneous and high-dimensional data is being a chall Estimation and feature selection in high dimensional mixtures of experts models (2) enging problem, both from modeling, and inference point of views, especially with the today’s big data phenomenon. This suggests new strategics, partiEstimation and feature selection in high dimensional mixtures of experts models (2)
cularly in advanced analyses going from density estimation to prediction, as well as the unsupervised classification, of many kinds of such data with Normandie UniversitéTHESEPour obtenir le diplôme de doctoratSpécialité MATHEMATIQUESPréparée au sein de 1'Université de Caen NormandieEstimation and F Estimation and feature selection in high dimensional mixtures of experts models (2) luding density estimation and clustering, and their elegant Mixtures-of-Experts (MoE) variety, which strengthen the link with supervised learning and hence deals furthermore with prediction from heterogeneous regressiontype data, and for classification. In a high-dimensional scenario, particularly f Estimation and feature selection in high dimensional mixtures of experts models (2) or data arising from a heterogeneous population, using such MoE models requires addressing modeling and estimation questions, since the state-of-the aEstimation and feature selection in high dimensional mixtures of experts models (2)
rt estimation methodologies arc limited.This thesis deals with the problem of modeling and estimation of high-dimensional MoE models, towards effectivNormandie UniversitéTHESEPour obtenir le diplôme de doctoratSpécialité MATHEMATIQUESPréparée au sein de 1'Université de Caen NormandieEstimation and F Estimation and feature selection in high dimensional mixtures of experts models (2) mum-likelihood estimation (MLE) of MoE models to overcome the limitations of standard methods, including MLE estimation with Expectation-Maximization (EM) algorithms, and to simultaneously perform feature selection so that sparse models are encouraged in such a high-dimensional setting. We first int Estimation and feature selection in high dimensional mixtures of experts models (2) roduce a mixture-of-experts* parameter estimation and variable selection methodology, based on £] (lasso) regularizations and the EM framework, for reEstimation and feature selection in high dimensional mixtures of experts models (2)
gression ami clustering suited to highdimensional contexts. Then, we extend the method to regularized mixture of experts models foldiscrete data, inclNormandie UniversitéTHESEPour obtenir le diplôme de doctoratSpécialité MATHEMATIQUESPréparée au sein de 1'Université de Caen NormandieEstimation and F Estimation and feature selection in high dimensional mixtures of experts models (2) gies enjoy the efficient monotone maximization of the optimized criterion, and unlike previous approaches, they do not rely on approximations on the penalty functions, avoid matrix inversion, ami exploit the efficiency of the coordinate ascent algorithm, particularly within the proximal Newton-based Estimation and feature selection in high dimensional mixtures of experts models (2) approach.Keywords: Mixture models; Mixture of Experts; Regularized Estimation; Feature Selection; Lasso; (J-regularization; Sparsity; EM algorithm: MEstimation and feature selection in high dimensional mixtures of experts models (2)
M Algorithm: Proximal-Newton; Coordinate Ascent: Clustering; Classification; Regression; Prediction.Normandie UniversitéTHESEPour obtenir le diplôme de doctoratSpécialité MATHEMATIQUESPréparée au sein de 1'Université de Caen NormandieEstimation and FNormandie UniversitéTHESEPour obtenir le diplôme de doctoratSpécialité MATHEMATIQUESPréparée au sein de 1'Université de Caen NormandieEstimation and FGọi ngay
Chat zalo
Facebook