High-dimensional Gaussian model selection on a Gaussian design

DOI10.1214/09-AIHP321zbMATH Open1191.62076arXiv0808.2152OpenAlexW2963498088MaRDI QIDQ985331FDOQ985331

Nicolas Verzelen

Publication date: 21 July 2010

Published in: Annales de l'Institut Henri Poincaré. Probabilités et Statistiques (Search for Journal in Brave)

Abstract: We consider the problem of estimating the conditional mean of a real Gaussian variable $olinebreakY=su{m}_{i=1}^{p}olinebreakhet{a}_i{X}_i+olinebreakepsilon$ where the vector of the covariates $(X_i{)}_{1leqileqp}$ follows a joint Gaussian distribution. This issue often occurs when one aims at estimating the graph or the distribution of a Gaussian graphical model. We introduce a general model selection procedure which is based on the minimization of a penalized least-squares type criterion. It handles a variety of problems such as ordered and complete variable selection, allows to incorporate some prior knowledge on the model and applies when the number of covariates $p$ is larger than the number of observations $n$. Moreover, it is shown to achieve a non-asymptotic oracle inequality independently of the correlation structure of the covariates. We also exhibit various minimax rates of estimation in the considered framework and hence derive adaptiveness properties of our procedure.

Full work available at URL: https://arxiv.org/abs/0808.2152

zbMATH Keywords

model selectionGaussian graphical modelslinear regressionoracle inequalitiesminimax rates of estimation

Mathematics Subject Classification ID

Nonparametric regression and quantile regression (62G08) Linear regression; mixed models (62J05) Inequalities; stochastic orderings (60E15)

Cites Work

• Title not available (Why is that?)
• Title not available (Why is that?)
• Title not available (Why is that?)
• Title not available (Why is that?)
• Title not available (Why is that?)
• Title not available (Why is that?)
• Gaussian Markov Random Fields
• Estimating the dimension of a model
• The Adaptive Lasso and Its Oracle Properties
• Some Comments on C P
• A new look at the statistical model identification
• Simultaneous analysis of Lasso and Dantzig selector
• The Dantzig selector: statistical estimation when \(p\) is much larger than \(n\). (With discussions and rejoinder).
• High-dimensional graphs and variable selection with the Lasso
• Learning Theory and Kernel Machines
• Tests for Gaussian graphical models
• Gaussian model selection with an unknown variance
• Sparsity oracle inequalities for the Lasso
• Gaussian model selection
• Local operator theory, random matrices and Banach spaces.
• Decoding by Linear Programming
• Adaptive estimation of a quadratic functional by model selection.
• Estimation of Gaussian graphs by model selection
• Model selection by resampling penalization
• Aggregation for Gaussian regression
• Minimum contrast estimators on sieves: Exponential bounds and rates of convergence
• Probabilistic Networks and Expert Systems
• Minimal penalties for Gaussian model selection
• Concentration inequalities and model selection. Ecole d'Eté de Probabilités de Saint-Flour XXXIII -- 2003.
• An optimal selection of regression variables
• Statistical predictor identification
• Near-ideal model selection by \(\ell _{1}\) minimization
• Information-Theoretic Limits on Sparsity Recovery in the High-Dimensional and Noisy Setting
• Power-law correlations, related models for long-range dependence and their simulation
• A New Lower Bound for Multiple Hypothesis Testing

Cited In (9)

• Penalized contrast estimation in functional linear models with circular data
• Title not available (Why is that?)
• Adaptive estimation of covariance matrices via Cholesky decomposition
• Minimax risks for sparse regressions: ultra-high dimensional phenomenons
• Adaptive estimation of linear functionals in functional linear models
• Model selection for Gaussian regression with random design
• Adaptive functional linear regression
• Gaussian model selection
• High-dimensional regression with unknown variance

Uses Software

• pcalg
• GMRFLib

This page was built for publication: High-dimensional Gaussian model selection on a Gaussian design

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q985331)