Screening Sinkhorn Algorithm for Regularized Optimal Transport

Abstract

We introduce in this paper a novel strategy for efficiently approximating the Sinkhorn distance between two discrete measures. After identifying neglectable components of the dual solution of the regularized Sinkhorn problem, we propose to screen those components by directly setting them at that value before entering the Sinkhorn problem. This allows us to solve a smaller Sinkhorn problem while ensuring approximation with provable guarantees. More formally, the approach is based on a new formulation of dual of Sinkhorn divergence problem and on the KKT optimality conditions of this problem, which enable identification of dual components to be screened. This new analysis leads to the Screenkhorn algorithm. We illustrate the efficiency of Screenkhorn on complex tasks such as dimensionality reduction and domain adaptation involving regularized optimal transport.

References in corpus (11)

• Learning with a Wasserstein Loss
• Near-linear time approximation algorithms for optimal transport via Sinkhorn iteration
• Safe Feature Elimination in Sparse Supervised Learning
• Amplitude and phase variation of point processes
• Wasserstein Discriminant Analysis
• Smooth and Sparse Optimal Transport
• On Efficient Optimal Transport: An Analysis of Greedy and Accelerated Mirror Descent Algorithms
• Multilevel Clustering via Wasserstein Means
• Massively scalable Sinkhorn distances via the Nyström method
• Greedy stochastic algorithms for entropy-regularized optimal transport problems
• Construction of Non-asymptotic Confidence Sets in 2-Wasserstein Space

Cited by in corpus (5)

• Accelerating the Sinkhorn algorithm for sparse multi-marginal optimal transport by fast Fourier transforms
• Flow-based Alignment Approaches for Probability Measures in Different Spaces
• CO-Optimal Transport
• A Novel Skip Orthogonal List for Dynamic Optimal Transport Problem
• Using Optimal Transport as Alignment Objective for fine-tuning Multilingual Contextualized Embeddings