Central limit theorems for entropy-regularized optimal transport on finite spaces and statistical applications

DOI10.1214/19-EJS1637zbMATH Open1454.62136arXiv1711.08947OpenAlexW2913699596MaRDI QIDQ2283574FDOQ2283574

Authors: Jérémie Bigot, Elsa Cazelles, Nicolas Papadakis 

Publication date: 3 January 2020

Published in: Electronic Journal of Statistics (Search for Journal in Brave)

Abstract: The notion of entropy-regularized optimal transport, also known as Sinkhorn divergence, has recently gained popularity in machine learning and statistics, as it makes feasible the use of smoothed optimal transportation distances for data analysis. The Sinkhorn divergence allows the fast computation of an entropically regularized Wasserstein distance between two probability distributions supported on a finite metric space of (possibly) high-dimension. For data sampled from one or two unknown probability distributions, we derive the distributional limits of the empirical Sinkhorn divergence and its centered version (Sinkhorn loss). We also propose a bootstrap procedure which allows to obtain new test statistics for measuring the discrepancies between multivariate probability distributions. Our work is inspired by the results of Sommerfeld and Munk (2016) on the asymptotic distribution of empirical Wasserstein distance on finite space using unregularized transportation costs. Incidentally we also analyze the asymptotic distribution of entropy-regularized Wasserstein distances when the regularization parameter tends to zero. Simulated and real datasets are used to illustrate our approach.

Full work available at URL: https://arxiv.org/abs/1711.08947

Recommendations

• An Improved Central Limit Theorem and Fast Convergence Rates for Entropic Transportation Costs
• Optimal transportation-entropy inequalities for several usual distributions on \(\mathbb R\)
• Empirical optimal transport on countable metric spaces: distributional limits and statistical applications
• Empirical regularized optimal transport: statistical theory and applications
• Asymptotic distribution and convergence rates of stochastic algorithms for entropic optimal transportation between probability measures
• Entropy inequalities for stable densities and strengthened central limit theorems
• Transport-entropy inequalities and deviation estimates for stochastic approximation schemes
• Entropic optimal transport: geometry and large deviations
• A note on the regularity of optimal-transport-based center-outward distribution and quantile functions
• Entropic regularization of continuous optimal transport problems

zbMATH Keywords

bootstraphypothesis testingcentral limit theoremoptimal transportmultivariate probability distributionsSinkhorn divergence

Mathematics Subject Classification ID

Computational methods for problems pertaining to statistics (62-08) Nonparametric hypothesis testing (62G10) Asymptotic properties of nonparametric inference (62G20) Nonparametric statistical resampling methods (62G09) Central limit and other weak theorems (60F05)

Cites Work

• Title not available (Why is that?)
• Weak convergence and empirical processes. With applications to statistics
• Title not available (Why is that?)
• Tests of goodness of fit based on the \(L_2\)-Wasserstein distance
• Wasserstein dictionary learning: optimal transport-based unsupervised nonlinear dictionary learning
• Convolutional Wasserstein distances: efficient optimal transportation on geometric domains
• Convergence of latent mixing measures in finite and infinite mixture models
• Title not available (Why is that?)
• Title not available (Why is that?)
• Title not available (Why is that?)
• Limit laws of the empirical Wasserstein distance: Gaussian distributions
• Asymptotics for \(L_2\) functionals of the empirical quantile process, with applications to tests of fit based on weighted Wasserstein distances
• On Hadamard differentiability in \(k\)-sample semiparametric models -- with applications to the assessment of structural relationships
• Inverses of \(2\times 2\) block matrices
• Geodesic PCA versus Log-PCA of Histograms in the Wasserstein Space
• Geodesic PCA in the Wasserstein space by convex PCA
• Central limit theorems for empirical transportation cost in general dimension
• Convex color image segmentation with optimal transport distances
• A smoothed dual approach for variational Wasserstein problems
• Inference for empirical Wasserstein distances on finite spaces
• Fast Discrete Distribution Clustering Using Wasserstein Barycenter With Sparse Support
• Inference on functionals under first order degeneracy
• Empirical regularized optimal transport: statistical theory and applications
• Asymptotic analysis of objectives based on Fisher information in active learning
• Data-driven regularization of Wasserstein barycenters with an application to multivariate density registration

Cited In (18)

• Asymptotics of discrete Schrödinger bridges via chaos decomposition
• Asymptotic distribution and convergence rates of stochastic algorithms for entropic optimal transportation between probability measures
• Limit distribution theory for smooth \(p\)-Wasserstein distances
• Scaling positive random matrices: concentration and asymptotic convergence
• Empirical optimal transport on countable metric spaces: distributional limits and statistical applications
• Projection-based techniques for high-dimensional optimal transport problems
• An Improved Central Limit Theorem and Fast Convergence Rates for Entropic Transportation Costs
• Empirical regularized optimal transport: statistical theory and applications
• The nested Sinkhorn divergence to learn the nested distance
• Gromov-Wasserstein distances: entropic regularization, duality and sample complexity
• Minimax estimation of smooth optimal transport maps
• Stochastic approximation versus sample average approximation for Wasserstein barycenters
• A central limit theorem for Lp transportation cost on the real line with application to fairness assessment in machine learning
• Limit theorems for entropic optimal transport maps and Sinkhorn divergence
• Uniform confidence band for optimal transport map on one-dimensional data
• Estimation of Wasserstein distances in the spiked transport model
• Statistical inference with regularized optimal transport
• Limit distributions and sensitivity analysis for empirical entropic optimal transport on countable spaces

Uses Software

• bootstrap
• Wasserstein GAN

This page was built for publication: Central limit theorems for entropy-regularized optimal transport on finite spaces and statistical applications

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2283574)