Debiased Sinkhorn barycenters - Inria - Institut national de recherche en sciences et technologies du numérique Access content directly
Conference Papers Year : 2020

Debiased Sinkhorn barycenters

Abstract

Entropy regularization in optimal transport (OT) has been the driver of many recent interests for Wasserstein metrics and barycenters in machine learning. It allows to keep the appealing geometrical properties of the unregularized Wasserstein distance while having a significantly lower complexity thanks to Sinkhorn's algorithm. However, entropy brings some inherent smoothing bias, resulting for example in blurred barycenters. This side effect has prompted an increasing temptation in the community to settle for a slower algorithm such as log-domain stabilized Sinkhorn which breaks the parallel structure that can be leveraged on GPUs, or even go back to unregularized OT. Here we show how this bias is tightly linked to the reference measure that defines the entropy regularizer and propose debiased Wasserstein barycenters that preserve the best of both worlds: fast Sinkhorn-like iterations without entropy smoothing. Theoretically, we prove that the entropic OT barycenter of univariate Gaussians is a Gaussian and quantify its variance bias. This result is obtained by extending the differentiability and convexity of entropic OT to sub-Gaussian measures with unbounded supports. Empirically, we illustrate the reduced blurring and the computational advantage on various applications.
Fichier principal
Vignette du fichier
2006.02575.pdf (1.75 Mo) Télécharger le fichier
Origin : Files produced by the author(s)

Dates and versions

hal-03063875 , version 1 (14-12-2020)

Identifiers

  • HAL Id : hal-03063875 , version 1

Cite

Hicham Janati, Marco Cuturi, Alexandre Gramfort. Debiased Sinkhorn barycenters. ICML 2020 - 37th International Conference on Machine Learning, Jul 2020, Vienna / Virtuel, Austria. ⟨hal-03063875⟩
85 View
54 Download

Share

Gmail Facebook X LinkedIn More