Sinkhorn Distances: Lightspeed Computation of Optimal Transportation Distances

Abstract

Optimal transportation distances are a fundamental family of parameterized distances for histograms. Despite their appealing theoretical properties, excellent performance in retrieval tasks and intuitive formulation, their computation involves the resolution of a linear program whose cost is prohibitive whenever the histograms' dimension exceeds a few hundreds. We propose in this work a new family of optimal transportation distances that look at transportation problems from a maximum-entropy perspective. We smooth the classical optimal transportation problem with an entropic regularization term, and show that the resulting optimum is also a distance which can be computed through Sinkhorn-Knopp's matrix scaling algorithm at a speed that is several orders of magnitude faster than that of transportation solvers. We also report improved performance over classical optimal transportation distances on the MNIST benchmark problem.

Cited by in corpus (30)

• Machine Learning Unifies the Modelling of Materials and Molecules
• Learning to Simulate Complex Physics with Graph Networks
• The Born Supremacy: Quantum Advantage and Training of an Ising Born Machine
• A Sparse Multi-Scale Algorithm for Dense Optimal Transport
• Cross-domain Activity Recognition via Substructural Optimal Transport
• Two-sample Test using Projected Wasserstein Distance
• Motif-Driven Contrastive Learning of Graph Representations
• Wasserstein GANs Work Because They Fail (to Approximate the Wasserstein Distance)
• Domain decomposition for entropy regularized optimal transport
• SSFL: Tackling Label Deficiency in Federated Learning via Personalized Self-Supervision
• Orlicz space regularization of continuous optimal transport problems
• Learning with Fenchel-Young Losses
• Cross-Trajectory Representation Learning for Zero-Shot Generalization in RL
• Minimizing relative entropy of path measures under marginal constraints
• Stochastic Adversarial Koopman Model for Dynamical Systems
• Nonparametric two-sample hypothesis testing for low-rank random graphs of differing sizes
• Two-sample Test with Kernel Projected Wasserstein Distance
• On Coupling Particle Filter Trajectories
• Fast block-coordinate Frank-Wolfe algorithm for semi-relaxed optimal transport
• An Exploration of Learnt Representations of W Jets
• GANs with Variational Entropy Regularizers: Applications in Mitigating the Mode-Collapse Issue
• Volume Preserving Image Segmentation with Entropic Regularization Optimal Transport and Its Applications in Deep Learning
• New Metrics Between Rational Spectra and their Connection to Optimal Transport
• Feedback Particle Filter With Stochastically Perturbed Innovation And Its Application to Dual Estimation
• Using Optimal Transport as Alignment Objective for fine-tuning Multilingual Contextualized Embeddings
• Permutation invariant networks to learn Wasserstein metrics
• SwAMP: Swapped Assignment of Multi-Modal Pairs for Cross-Modal Retrieval
• Transport based Graph Kernels
• Distributionally-Constrained Policy Optimization via Unbalanced Optimal Transport
• Learning Semidefinite Regularizers