ON THE SAMPLE COMPLEXITY OF ENTROPIC OPTIMAL TRANSPORT

Research output: Contribution to journalArticlepeer-review

Abstract

We study the sample complexity of entropic optimal transport in high dimensions using computationally efficient plug-in estimators. We significantly advance the state of the art by establishing dimension-free, parametric rates for estimating various quantities of interest, including the entropic regression function, which is a natural analog to the optimal transport map. As an application, we propose a practical model for transfer learning based on entropic optimal transport and establish parametric rates of convergence for nonparametric regression and classification.

Original languageEnglish
Pages (from-to)61-90
Number of pages30
JournalAnnals of Statistics
Volume53
Issue number1
DOIs
Publication statusPublished - 1 Feb 2025
Externally publishedYes

Keywords

  • Optimal transport
  • Schrödinger bridge
  • entropic regularization
  • statistical rates

Fingerprint

Dive into the research topics of 'ON THE SAMPLE COMPLEXITY OF ENTROPIC OPTIMAL TRANSPORT'. Together they form a unique fingerprint.

Cite this