Union Support Recovery in High-Dimensional Multivariate Regression

Abstract

In the problem of multivariate regression, a K-dimensional response vector is regressed upon a common set of p covariates, with a matrix of regression coefficients. We study the behavior of the group Lasso using l1/l2 regularization for the union support problem, meaning that the set of s rows for which B* is non-zero is recovered exactly. Studying this problem under high-dimensional scaling, we show that group Lasso recovers the exact row pattern with high probability over the random design and noise for scalings of such that the sample complexity parameter exceeds a critical threshold. Here n is the sample size, p is the ambient dimension of the regression model, s is the number of non-zero rows, and B* is a sparsity-overlap function that measures a combination of the sparsities and overlaps of the K-regression coefficient vectors that constitute the model.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Aug 01, 2008
Accession Number
ADA488564

Entities

People

  • Guillaume Obozinski
  • Martin J. Wainwright
  • Michael I. Jordan

Organizations

  • University of California, Berkeley

Tags

Communities of Interest

  • C4I
  • Energy and Power Technologies
  • Materials and Manufacturing Processes

DTIC Thesaurus Topics

  • Algorithms
  • Computational Complexity
  • Computer Science
  • Consistency
  • Covariance
  • Electrical Engineering
  • Equations
  • Information Processing
  • Information Science
  • Machine Learning
  • Numerical Analysis
  • Optimization
  • Probability
  • Recovery
  • Statistical Analysis
  • Statistics
  • Theorems

Fields of Study

  • Computer science

Readers

  • Graph Algorithms and Convex Optimization.
  • Neural Network Machine Learning.
  • Regression Analysis.