Statistical Aspects of Wasserstein Distances

Abstract

Wasserstein distances are metrics on probability distributions inspired by the problem of optimal mass transportation. Roughly speaking, they measure the minimal effort required to reconfigure the probability mass of one distribution in order to recover the other distribution. They are ubiquitous in mathematics, with a long history that has seen them catalyse core developments in analysis, optimization, and probability. Beyond their intrinsic mathematical richness, they possess attractive features that make them a versatile tool for the statistician: they can be used to derive weak convergence and convergence of moments, and can be easily bounded; they are well-adapted to quantify a natural notion of perturbation of a probability distribution; and they seamlessly incorporate the geometry of the domain of the distributions in question, thus being useful for contrasting complex objects. Consequently, they frequently appear in the development of statistical theory and inferential methodology, and have recently become an object of inference in themselves. In this review, we provide a snapshot of the main concepts involved in Wasserstein distances and optimal transportation, and a succinct overview of some of their many statistical aspects.

arXiv commentOfficial version available at https://www.annualreviews.org/doi/full/10.1146/annurev-statistics-030718-104938

References in corpus (16)

• Fast Computation of Wasserstein Barycenters
• Functional Data Analysis of Amplitude and Phase Variation
• Spectral gaps for a Metropolis-Hastings algorithm in infinite dimensions
• Amplitude and phase variation of point processes
• Fast transport optimization for Monge costs on the circle
• Learning Probability Measures with respect to Optimal Transport Metrics
• Error bounds for Metropolis-Hastings algorithms applied to perturbations of Gaussian measures in high dimensions
• On the convergence to equilibrium of Kac's random walk on matrices
• On parameter estimation with the Wasserstein distance
• Characterization of barycenters in the Wasserstein space by averaging optimal transport maps
• A geometric study of Wasserstein spaces: ultrametrics
• Convergence rates for empirical barycenters in metric spaces: curvature, convexity and extendible geodesics
• Penalized Barycenters in the Wasserstein Space
• Stein's method and Poisson process approximation for a class of Wasserstein metrics
• Central Limit Theorem for empirical transportation cost in general dimension
• Minimax convergence rate for estimating the Wasserstein barycenter of random measures on the real line

Cited by in corpus (73)

• Metrics for Graph Comparison: A Practitioner's Guide
• Atom-Density Representations for Machine Learning
• Approximate Bayesian computation with the Wasserstein distance
• Glucodensities: a new representation of glucose profiles using distributional data analysis
• Generative diffusion in very large dimensions
• Wasserstein barycenters are NP-hard to compute
• DAOT: Domain-Agnostically Aligned Optimal Transport for Domain-Adaptive Crowd Counting
• Minimax Confidence Intervals for the Sliced Wasserstein Distance
• An Optimal Transport Approach to Personalized Federated Learning
• Understanding spatial propagation using metric geometry with application to the spread of COVID-19 in the United States
• Automated Grain Boundary Detection for Bright-Field Transmission Electron Microscopy Images via U-Net
• Gaussian Belief Trees for Chance Constrained Asymptotically Optimal Motion Planning
• Knowledge Distillation in Federated Learning: a Survey on Long Lasting Challenges and New Solutions
• Topological Data Analysis of Human Brain Networks Through Order Statistics
• Visual Analysis of Large Multivariate Scattered Data using Clustering and Probabilistic Summaries
• Probabilistic surrogate modeling of offshore wind-turbine loads with chained Gaussian processes
• Empirical geodesic graphs and CAT(k) metrics for data analysis
• 1-Wasserstein Distance on the Standard Simplex
• Information-geometry of physics-informed statistical manifolds and its use in data assimilation
• Martingale Methods for Sequential Estimation of Convex Functionals and Divergences
• Earth mover's distance as a measure of CP violation
• Center-Outward Distribution Functions, Quantiles, Ranks, and Signs in ${\mathbb{R}}^d$
• Smooth $p$-Wasserstein Distance: Structure, Empirical Approximation, and Statistical Applications
• Dynamical Mode Recognition of Coupled Flame Oscillators by Supervised and Unsupervised Learning Approaches
• Nonlinear projection-based model order reduction with machine learning regression for closure error modeling in the latent space
• Sufficient dimension reduction for classification using principal optimal transport direction
• Entropic-Wasserstein barycenters: PDE characterization, regularity and CLT
• Limit Laws for Empirical Optimal Solutions in Stochastic Linear Programs
• Wasserstein barycenters can be computed in polynomial time in fixed dimension
• Order $p$ quantum Wasserstein distances from couplings
• Gaussian-Smooth Optimal Transport: Metric Structure and Statistical Efficiency
• Averaging on the Bures-Wasserstein manifold: dimension-free convergence of gradient descent
• On Projection Robust Optimal Transport: Sample Complexity and Model Misspecification
• Conditional Wasserstein Barycenters and Interpolation/Extrapolation of Distributions
• CLAP. I. Resolving miscalibration for deep learning-based galaxy photometric redshift estimation
• Improving Neural Topic Models with Wasserstein Knowledge Distillation
• Nonlinear Inverse Optimal Transport: Identifiability of the Transport Cost from its Marginals and Optimal Values
• Wavelet-Based CSI Reconstruction for Improved Wireless Security Through Channel Reciprocity
• Novelty Detection via Robust Variational Autoencoding
• Refined climatologies of future precipitation over High Mountain Asia using probabilistic ensemble learning
• Consensus group decision making under model uncertainty with a view towards environmental policy making
• Automated regime classification in multidimensional time series data using sliced Wasserstein k-means clustering
• Regularity as Regularization: Smooth and Strongly Convex Brenier Potentials in Optimal Transport
• Intrinsic Sliced Wasserstein Distances for Comparing Collections of Probability Distributions on Manifolds and Graphs
• Exploring Predictive States via Cantor Embeddings and Wasserstein Distance
• Dimensionality Reduction and Dynamical Mode Recognition of Circular Arrays of Flame Oscillators Using Deep Neural Network
• Dynamical Wasserstein Barycenters for Time-series Modeling
• Inferring traits of hyperuniformity from local structures via persistent homology
• Pioneering High-Speed Pulsar Parameter Estimation Using Convolutional Neural Networks
• DP2Unlearning: An Efficient and Guaranteed Unlearning Framework for LLMs
• Stochastic Approximation versus Sample Average Approximation for population Wasserstein barycenters
• Fisher-Pitman permutation tests based on nonparametric Poisson mixtures with application to single cell genomics
• Probabilistic Outlier Detection and Generation
• Cutoff thermalization for Ornstein-Uhlenbeck systems with small Lévy noise in the Wasserstein distance
• Asymptotic Stochastic Comparison of Random Processes
• Monitoring road infrastructures from satellite images in Greater Maputo
• Causal Inference on Distribution Functions
• Distributionally Robust Model Order Reduction for Linear Systems
• Value-Enriched Population Synthesis: Integrating a Motivational Layer
• Quantum Process Tomography with Digital Twins of Error Matrices
• On-Policy Model Errors in Reinforcement Learning
• A Graph Gaussian Embedding Method for Predicting Alzheimer's Disease Progression with MEG Brain Networks
• Approximate Heavy Tails in Offline (Multi-Pass) Stochastic Gradient Descent
• Fast variable selection for distributional regression with application to continuous glucose monitoring data
• Expected value and a Cayley-Menger type formula for the generalized earth mover's distance
• A Pseudo-Metric between Probability Distributions based on Depth-Trimmed Regions
• Invariant Measures in Time-Delay Coordinates for Unique Dynamical System Identification
• Riemannian-geometric generalizations of quantum fidelities and Bures-Wasserstein distance
• Compatibility-based Bayesian inference: asymptotic limits, variational principles, and Wasserstein geometry
• The coupling method in extreme value theory
• Reweighting samples under covariate shift using a Wasserstein distance criterion
• Permutation invariant networks to learn Wasserstein metrics
• Statistical techniques to estimate the SARS-CoV-2 infection fatality rate