Learning Markov Network Structure using Brownian Distance Covariance

TL;DR

This paper introduces a non-parametric approach using Brownian distance covariance to learn the structure of undirected Markov networks from data, effective even in high-dimensional scenarios.

Contribution

It presents a novel method for structure learning of Markov networks leveraging Brownian distance covariance, suitable for high-dimensional data.

Findings

Effective in high-dimensional settings

Accurately estimates conditional independences

Applicable to various types of data

Abstract

In this paper, we present a simple non-parametric method for learning the structure of undirected graphs from data that drawn from an underlying unknown distribution. We propose to use Brownian distance covariance to estimate the conditional independences between the random variables and encodes pairwise Markov graph. This framework can be applied in high-dimensional setting, where the number of parameters much be larger than the sample size.