Optimal Transport losses and Sinkhorn algorithm with general convex regularization

Abstract

We introduce a new class of convex-regularized Optimal Transport losses, which generalizes the classical Entropy-regularization of Optimal Transport and Sinkhorn divergences, and propose a generalized Sinkhorn algorithm. Our framework unifies many regularizations and numerical methods previously appeared in the literature. We show the existence of the maximizer for the dual problem, complementary slackness conditions, providing a complete characterization of solutions for such class of variational problems. As a consequence, we study structural properties of these losses, including continuity, differentiability and provide explicit formulas for the its gradient. Finally, we provide theoretical guarantees of convergences and stability of the generalized Sinkhorn algorithm, even in the continuous setting. The techniques developed here are directly applicable also to study Wasserstein barycenters or, more generally, multi-marginal problems.

References in corpus (13)

• Fast Computation of Wasserstein Barycenters
• Learning with a Wasserstein Loss
• Stochastic Optimization for Large-scale Optimal Transport
• Sample Complexity of Sinkhorn divergences
• Interpolating between Optimal Transport and MMD using Sinkhorn Divergences
• Smooth and Sparse Optimal Transport
• Scaling Algorithms for Unbalanced Transport Problems
• Differential Properties of Sinkhorn Approximation for Learning with Wasserstein Distance
• A review of matrix scaling and Sinkhorn's normal form for matrices and positive maps
• Principal Geodesic Analysis for Probability Measures under the Optimal Transport Metric
• On the Complexity of Approximating Multimarginal Optimal Transport
• Generalized conditional gradient: analysis of convergence and applications
• Orlicz-space regularization for optimal transport and algorithms for quadratic regularization

Cited by in corpus (5)

• Orlicz space regularization of continuous optimal transport problems
• Sampling-Based Methods for Multi-Block Optimization Problems over Transport Polytopes
• Cortically based optimal transport
• Estimating 2-Sinkhorn Divergence between Gaussian Processes from Finite-Dimensional Marginals
• Optimal transport with $f$-divergence regularization and generalized Sinkhorn algorithm