k-Sample inference via Multimarginal Optimal Transport

TL;DR

This paper introduces a multimarginal optimal transport method for comparing multiple measures, deriving asymptotic distributions for hypothesis testing, and demonstrating its effectiveness on synthetic and real datasets including cancer data.

Contribution

It develops a novel $MOT$ framework for $k$-sample testing, providing asymptotic theory, bootstrap estimation, and a low complexity approximation method.

Findings

Effective $k$-sample test with theoretical guarantees.

Bootstrap-based distribution estimation improves accuracy.

Successful application to cancer dataset analysis.

Abstract

This paper proposes a Multimarginal Optimal Transport ($MOT$) approach for simultaneously comparing $k\geq 2$ measures supported on finite subsets of $\mathbb{R}^d$, $d \geq 1$. We derive asymptotic distributions of the optimal value of the empirical $MOT$ program under the null hypothesis that all $k$ measures are same, and the alternative hypothesis that at least two measures are different. We use these results to construct the test of the null hypothesis and provide consistency and power guarantees of this $k$-sample test. We consistently estimate asymptotic distributions using bootstrap, and propose a low complexity linear program to approximate the test cut-off. We demonstrate the advantages of our approach on synthetic and real datasets, including the real data on cancers in the United States in 2004 - 2020.