Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Interpolating between Optimal Transport and MMD using Sinkhorn Divergences

About

Comparing probability distributions is a fundamental problem in data sciences. Simple norms and divergences such as the total variation and the relative entropy only compare densities in a point-wise manner and fail to capture the geometric nature of the problem. In sharp contrast, Maximum Mean Discrepancies (MMD) and Optimal Transport distances (OT) are two classes of distances between measures that take into account the geometry of the underlying space and metrize the convergence in law. This paper studies the Sinkhorn divergences, a family of geometric divergences that interpolates between MMD and OT. Relying on a new notion of geometric entropy, we provide theoretical guarantees for these divergences: positivity, convexity and metrization of the convergence in law. On the practical side, we detail a numerical scheme that enables the large scale application of these divergences for machine learning: on the GPU, gradients of the Sinkhorn loss can be computed for batches of a million samples.

Jean Feydy, Thibault S\'ejourn\'e, Fran\c{c}ois-Xavier Vialard, Shun-ichi Amari, Alain Trouv\'e, Gabriel Peyr\'e• 2018

Related benchmarks

TaskDatasetResultRank
Metric Correlation AnalysisCIFAR-10 and LSUN Timesteps 0-999 Averaged over 5 runs across 36 model configurations (Full trajectory)
L2 Norm of Mean (||mu_t||_2)0.0831
8
Metric Correlation AnalysisCIFAR-10 and LSUN Timesteps 900-999 Initial denoising stages
L2 Norm (mu_t)0.0399
8
Showing 2 of 2 rows

Other info

Follow for update