Using Monte Carlo Tree Search to Calculate Mutual Information in High Dimensions

TL;DR

This paper introduces a novel approach combining Monte Carlo Tree Search with existing methods to improve the accuracy of mutual information estimation in high-dimensional, noisy data scenarios.

Contribution

It presents a modified Monte Carlo Tree Search technique that significantly enhances mutual information calculation accuracy in complex, high-dimensional settings.

Findings

The method outperforms standard techniques in high-dimensional cases.

It provides accurate mutual information estimates where traditional methods fail.

Software implementation is publicly available for reproducibility.

Abstract

Mutual information is an important measure of the dependence among variables. It has become widely used in statistics, machine learning, biology, etc. However, the standard techniques for estimating it often perform poorly in higher dimensions or with noisy variables. Here we significantly improve one of the standard methods for calculating mutual information by combining it with a modified Monte Carlo Tree Search. We present results which show that our method gives accurate results where the standard methods fail. We also describe the software implementation of our method and give details on the publicly-available code.