A Pathwise Algorithm for Covariance Selection

TL;DR

This paper introduces an efficient pathwise algorithm for covariance selection that estimates sparse inverse covariance matrices by balancing likelihood and sparsity through regularization.

Contribution

It presents a novel, efficient algorithm to compute the entire regularization path for covariance selection problems, improving computational feasibility.

Findings

Algorithm efficiently computes regularization paths

Enables better control over sparsity in covariance estimation

Improves computational speed for high-dimensional problems

Abstract

Covariance selection seeks to estimate a covariance matrix by maximum likelihood while restricting the number of nonzero inverse covariance matrix coefficients. A single penalty parameter usually controls the tradeoff between log likelihood and sparsity in the inverse matrix. We describe an efficient algorithm for computing a full regularization path of solutions to this problem.