Sliced Optimal Transport Plans

TL;DR

This paper rigorously analyzes and generalizes methods for computing transport plans using sliced optimal transport, extending their applicability to generic probability measures and demonstrating their practical effectiveness in image processing tasks.

Contribution

It introduces the Pivot Sliced Discrepancy and generalizes Expected Sliced plans for generic measures, providing theoretical insights and practical algorithms.

Findings

Pivot Sliced Discrepancy is a semi-metric with a constrained Kantorovich relation.

Generalized sliced plans work effectively for arbitrary probability measures.

Numerical experiments validate the practical relevance of the proposed methods.

Abstract

Since the introduction of the Sliced Wasserstein distance in the literature, its simplicity and efficiency have made it one of the most interesting surrogate for the Wasserstein distance in image processing and machine learning. However, its inability to produce transport plans limits its practical use to applications where only a distance is necessary. Several heuristics have been proposed in the recent years to address this limitation when the probability measures are discrete. In this paper, we propose to study these different propositions by redefining and analysing them rigorously for generic probability measures. Leveraging the $\nu$-based Wasserstein distance and generalised geodesics, we introduce and study the Pivot Sliced Discrepancy, inspired by a recent work by Mahey et al.. We demonstrate its semi-metric properties and its relation to a constrained Kantorovich formulation. In the same way, we generalise and study the recent Expected Sliced plans introduced by Liu et al. for completely generic measures. Our theoretical contributions are supported by numerical experiments on synthetic and real datasets, including colour transfer and shape registration, evaluating the practical relevance of these different solutions.