Chain Rule Optimal Transport

DOI10.1007/978-3-030-65459-7_8MaRDI QIDQ5153518zbMATH OpenOpenAlexFDO

Authors Ke Sun, Frank Nielsen

Publication date 30 September 2021

Published in Signals and Communication Technology (Search for Journal in Brave)

Full work available at URL https://arxiv.org/abs/1812.08113

zbMATH Keywords

\(f\)-divergencesBregman divergenceinformation geometryJensen-Shannon divergencejoint convexityoptimal transportRényi divergencestatistical mixturestotal variation divergenceWasserstein distances

Mathematics Subject Classification ID

Optimal transportation (49Q22)

Abstract: We define a novel class of distances between statistical multivariate distributions by modeling an optimal transport problem on their marginals with respect to a ground distance defined on their conditionals. These new distances are metrics whenever the ground distance between the marginals is a metric, generalize both the Wasserstein distances between discrete measures and a recently introduced metric distance between statistical mixtures, and provide an upper bound for jointly convex distances between statistical mixtures. By entropic regularization of the optimal transport, we obtain a fast differentiable Sinkhorn-type distance. We experimentally evaluate our new family of distances by quantifying the upper bounds of several jointly convex distances between statistical mixtures, and by proposing a novel efficient method to learn Gaussian mixture models (GMMs) by simplifying kernel density estimators with respect to our distance. Our GMM learning technique experimentally improves significantly over the EM implementation of { t sklearn} on the { t MNIST} and { t Fashion MNIST} datasets.

Recommendations

• Constrained optimal transport
• Optimal Transport to a Variety
• Optimal transport and curvature
• The optimal partial transport problem
• Optimal transport between random measures
• Optimal transportation for the determinant
• Optimal transport on discrete domains
• Optimal Transport Over a Linear Dynamical System
• Optimal transport from a point-like source
• The directional optimal transport

Cites work

• A closed-form expression for the Sharma-Mittal entropy of exponential families
• A new class of metric divergences on probability spaces and its applicability in statistics
• Bayesian prediction based on a class of shrinkage priors for location-scale models
• Combinatorial optimization. Theory and algorithms
• Convex functions. Constructions, characterizations and counterexamples
• scientific article; zbMATH DE number 1579275 (Why is no real title available?)
• scientific article; zbMATH DE number 3567782 (Why is no real title available?)
• scientific article; zbMATH DE number 7370595 (Why is no real title available?)
• scientific article; zbMATH DE number 3429908 (Why is no real title available?)
• Information geometry and its applications
• Joint and separate convexity of the Bregman distance.
• Learning mixtures by simplifying kernel density estimators
• On a Formula for the L2 Wasserstein Metric between Measures on Euclidean and Hilbert Spaces
• On the joint convexity of the Bregman divergence of matrices
• On the translocation of masses
• Optimal transport for applied mathematicians. Calculus of variations, PDEs, and modeling
• Probabilistic distances between finite-state finite-alphabet hidden Markov models
• Properties of a Generalized Divergence Related to Tsallis Generalized Divergence
• Rényi Divergence and Kullback-Leibler Divergence
• Scikit-learn: machine learning in Python
• Sliced and Radon Wasserstein barycenters of measures
• Stability of RCD condition under concentration topology
• The earth mover's distance as a metric for image retrieval
• The Frechet distance between multivariate normal distributions
• The geometry of optimal transportation
• The role of perspective functions in convexity, polyconvexity, rank-one convexity and separate convexity
• The statistical Minkowski distances: closed-form formula for Gaussian mixture models
• The Wasserstein distance and approximation theorems
• Wasserstein geometry of Gaussian measures

Cited in

(7)

• Ground metric learning on graphs
• Optimal transport with some directed distances
• Radon-Sobolev variational auto-encoders
• Gaussian mixtures closest to a given measure via optimal transport
• The Wasserstein space of stochastic processes
• Denseness of biadapted Monge mappings
• A probabilistic view on the adapted Wasserstein distance

Describes a project that uses

Uses Software

• Scikit
• Fashion-MNIST
• MNIST
• TensorFlow

This page was built for publication: Chain Rule Optimal Transport

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q5153518)