The log-linear group-lasso estimator and its asymptotic properties

DOI10.3150/11-BEJ364zbMATH Open1243.62107arXiv0709.3526MaRDI QIDQ442085FDOQ442085

Authors: Yuval Nardi, Alessandro Rinaldo 

Publication date: 9 August 2012

Published in: Bernoulli (Search for Journal in Brave)

Abstract: We define the group-lasso estimator for the natural parameters of the exponential families of distributions representing hierarchical log-linear models under multinomial sampling scheme. Such estimator arises as the solution of a convex penalized likelihood optimization problem based on the group-lasso penalty. We illustrate how it is possible to construct an estimator of the underlying log-linear model using the blocks of nonzero coefficients recovered by the group-lasso procedure. We investigate the asymptotic properties of the group-lasso estimator as a model selection method in a double-asymptotic framework, in which both the sample size and the model complexity grow simultaneously. We provide conditions guaranteeing that the group-lasso estimator is model selection consistent, in the sense that, with overwhelming probability as the sample size increases, it correctly identifies all the sets of nonzero interactions among the variables. Provided the sequences of true underlying models is sparse enough, recovery is possible even if the number of cells grows larger than the sample size. Finally, we derive some central limit type of results for the log-linear group-lasso estimator.

Full work available at URL: https://arxiv.org/abs/0709.3526

Recommendations

• On the asymptotic properties of the group lasso estimator for linear models
• The Group Lasso for Logistic Regression
• Adaptive group Lasso for high-dimensional generalized linear models
• Consistent group selection in high-dimensional linear regression
• Adaptive lasso for generalized linear models with a diverging number of parameters

zbMATH Keywords

model selectionconsistency

Mathematics Subject Classification ID

Asymptotic properties of parametric estimators (62F12) Generalized linear models (logistic models) (62J12) Central limit and other weak theorems (60F05)

Cites Work

• Introduction to Graphical Modelling
• Discrete Multivariate Analysis Theory and Practice
• High-dimensional generalized linear models and the lasso
• High-dimensional graphs and variable selection with the Lasso
• Title not available (Why is that?)
• Sharp Thresholds for High-Dimensional and Noisy Sparsity Recovery Using $\ell _{1}$-Constrained Quadratic Programming (Lasso)
• Model Selection and Estimation in Regression with Grouped Variables
• A unified framework for high-dimensional analysis of \(M\)-estimators with decomposable regularizers
• Persistene in high-dimensional linear predictor-selection and the virtue of overparametrization
• On the dependence of the Berry-Esseen bound on dimension
• Three centuries of categorical data analysis: Log-linear models and maximum likelihood estima\-tion
• The Group Lasso for Logistic Regression
• Title not available (Why is that?)
• Title not available (Why is that?)
• The mode oriented stochastic search (MOSS) algorithm for log-linear models with conjugate priors
• High-dimensional Ising model selection using \(\ell _{1}\)-regularized logistic regression
• On the geometry of discrete exponential families with application to exponential random graph models
• Title not available (Why is that?)
• Nonconcave penalized likelihood with a diverging number of parameters.
• Asymptotic behavior of likelihood methods for exponential families when the number of parameters tends to infinity
• Goodness-of-fit statistics for discrete multivariate data
• On modified logarithmic Sobolev inequalities for Bernoulli and Poisson measures
• On the asymptotic properties of the group lasso estimator for linear models
• Title not available (Why is that?)
• Best subset selection, persistence in high-dimensional statistical learning and optimization under \(l_1\) constraint
• Asymptotic normality of posterior distributions for exponential families when the number of parameters tends to infinity.
• Theory of statistics
• Structured variable selection and estimation
• Markov fields and log-linear interaction models for contingency tables
• On the central limit theorem in \(R^ p\) when p\(\rightarrow \infty\)
• Central limit theorems for multinomial sums
• Title not available (Why is that?)
• Normal approximations to sums of scores based on occupancy numbers

Cited In (14)

• Model selection and local geometry
• Bayesian inference on group differences in multivariate categorical data
• Asymptotical efficiency of MLE with grouped data
• Group variable selection for relative error regression
• On the asymptotic properties of the group lasso estimator for linear models
• The Group Lasso for Logistic Regression
• Loglinear model selection and human mobility
• A lasso for hierarchical interactions
• Group variable selection and estimation in the Tobit censored response model
• Maximum likelihood estimation in log-linear models
• Composite mixture of log-linear models with application to psychiatric studies
• Autoregressive process modeling via the Lasso procedure
• Estimator augmentation with applications in high-dimensional group inference
• Coordinate descent based hierarchical interactive Lasso penalized logistic regression and its application to classification problems

Uses Software

• MIM

This page was built for publication: The log-linear group-lasso estimator and its asymptotic properties

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q442085)