Back to search

Cornell University

Computational and Statistical Properties of Optimal Transport-Based Distances

Abstract

dc:description.abstract

Over the past two decades, the statistical and computational properties of optimal transport (OT) have been systematically studied, driven in part by OT's broad applicability across data science, statistics, economics, physics, and other fields. This body of work has identified the curse of dimensionality inherent in statistical estimation with OT distances and has led the development of computationally and statistically efficient regularizations of OT. The first contribution of this thesis is a general framework for deriving limit distributions and other asymptotic properties of empirical regularized OT distances. While OT distances enable a natural comparison between distributions on a common space, comparing datasets of different types -- such as text and images -- requires specifying an ad hoc cost function, which may fail to capture a meaningful correspondence between data points. To address this issue, Gromov-Wasserstein (GW) distances have been proposed, enabling a comparison of metric measure spaces based on their intrinsic structure. As a result, GW distances have found widespread use in applications involving heterogeneous data. The second contribution of this thesis is to establish the first limit laws for empirical GW distances and to derive consistent resampling schemes. Given the broad applicability of GW distances, their computation is of a particular interest. However, existing algorithms for computing regularized GW distances lack comprehensive convergence rate guarantees. To address this gap, the final contribution of this thesis is to introduce the first algorithms for computing regularized GW distances subject to formal non-asymptotic convergence rate guarantees.

Degree

thesis:*
Name thesis:degree_name
Ph. D., Applied Mathematics
Level thesis:degree_level
Doctor of Philosophy
Discipline thesis:degree_discipline
Applied Mathematics
Grantor
Cornell University
Year dc:date.issued
2025

Author and committee

dc:creator, dc:contributor.*
Author dc:creator
  • Rioux, Gabriel
Chair dc:contributor.chair
  • Goldfeld, Ziv
Committee members dc:contributor.committeemember
  • Kato, Kengo
  • Yang, Yunan

Subjects

dc:subject × 6

Rights

dc:rights
Statement dc:rights
  • Attribution 4.0 International
Language dc:language.iso
en

Identifiers

dc:identifier.*
Dc Identifier Bibid
17422435
Dc Identifier Other
ProQuest Submission ID: 15086
ProQuest Publication ID: 32121277

Chain of custody

source
Harvested from
Cornell University
Base URL
ecommons.cornell.edu/server/oai/request
Last updated
2026-11-19
Source record
OAI-PMH GetRecord
citation

Rioux, Gabriel. Computational and Statistical Properties of Optimal Transport-Based Distances. Doctor of Philosophy thesis, Cornell University, 2025. https://hdl.handle.net/1813/120780