Sliced Orlicz-Wasserstein

2026-09-26

AI Summary

This study addresses the limited flexibility of the sliced Wasserstein distance and the computational inefficiency of the Orlicz-Wasserstein distance by introducing Orlicz functions into the slicing framework for the first time, proposing a sliced Orlicz-Wasserstein distance based on the Luxemburg norm. Theoretically, its metric properties, topological equivalence, and minimax optimality of estimation are rigorously established. Algorithmically, efficient approximation is achieved via Monte Carlo estimation combined with binary search. Experiments demonstrate that the proposed method preserves superior computational efficiency over the standard Orlicz-Wasserstein distance while exhibiting stronger capability in capturing distributional discrepancies compared to the conventional sliced Wasserstein distance in two-sample testing and generative model evaluation tasks.

Abstract

We propose sliced Orlicz-Wasserstein (SOW) distance which is a generalization of sliced Wasserstein (SW) distance. SOW replaces the $L^p$ norm in SW with a Luxemburg norm cost induced by an Orlicz function $\phi$. First, we prove that SOW distance is a metric on the space of measures with finite Orlicz norm, and show that it recovers the SW distance when the Orlicz function is $\phi(x)=x^p$. Next, we derive the topological properties of the SOW distance. In particular, we show that convergence under SOW implies weak convergence, and the converse is true under the compact support condition. We then present the theoretical results for estimating the SOW distance. We derive sample complexity for both the distance itself and the powered functional of the distance, and prove their minimax optimality. In addition, we discuss the computational algorithm for approximating the SOW distance by Monte-Carlo estimation and bisection search, as well as the associated approximation error and computational complexity analysis. Our experimental results reveal the superior computational efficiency of SOW compared with Orlicz-Wasserstein (OW) distance. Also, in the experiments, we demonstrate the favorable flexibility of SOW distance over SW in detecting differences between distributions by comparing their performance in two-sample tests and evaluating generative models on image datasets.

Problem

• Sliced Wasserstein distance
• Orlicz-Wasserstein distance
• probability metric
• distribution comparison
• generative model evaluation

Innovation

• Sliced Orlicz-Wasserstein distance
• Luxemburg norm
• sample complexity
• Monte-Carlo estimation
• generative model evaluation