Ricci curvature for parametric statistics via optimal transport

DOI10.1007/S41884-020-00026-2zbMATH Open1450.62010arXiv1807.07095OpenAlexW3004377896WikidataQ125748146 ScholiaQ125748146MaRDI QIDQ2006199FDOQ2006199

Authors: Wuchen Li, Guido Montúfar 

Publication date: 8 October 2020

Published in: Information Geometry (Search for Journal in Brave)

Abstract: We elaborate the notion of a Ricci curvature lower bound for parametrized statistical models. Following the seminal ideas of Lott-Strum-Villani, we define this notion based on the geodesic convexity of the Kullback-Leibler divergence in a Wasserstein statistical manifold, that is, a manifold of probability distributions endowed with a Wasserstein metric tensor structure. Within these definitions, the Ricci curvature is related to both, information geometry and Wasserstein geometry. These definitions allow us to formulate bounds on the convergence rate of Wasserstein gradient flows and information functional inequalities in parameter space. We discuss examples of Ricci curvature lower bounds and convergence rates in exponential family models.

Full work available at URL: https://arxiv.org/abs/1807.07095

Recommendations

• Ricci curvature for metric-measure spaces via optimal transport
• Optimal transport and Ricci curvature
• scientific article
• scientific article; zbMATH DE number 1053583
• Diffusion, optimal transport and Ricci curvature for metric measure space
• Ricci curvature, entropy, and optimal transport
• Statistics on Riemannian manifolds: asymptotic distribution and curvature
• Diffusion, optimal transport and Ricci curvature
• Optimal transport and Ricci curvature in Finsler geometry
• Embedding optimal transports in statistical manifolds

zbMATH Keywords

Ricci curvatureinformation projectionWasserstein statistical manifoldFokker-Planck equation on parameter space

Mathematics Subject Classification ID

Statistics on manifolds (62R30) Measures of information, entropy (94A17) Methods of local Riemannian geometry (53B21) Fokker-Planck equations (35Q84) Differential geometric aspects of statistical manifolds and information geometry (53B12) Information geometry (statistical aspects) (62B11)

Cites Work

• Optimal Transport
• A fast scoring algorithm for maximum likelihood estimation in unbalanced mixed models with nested random effects
• A computational fluid mechanics solution to the Monge-Kantorovich mass transfer problem
• Ricci curvature for metric-measure spaces via optimal transport
• On the geometry of metric measure spaces. I
• THE GEOMETRY OF DISSIPATIVE EVOLUTION EQUATIONS: THE POROUS MEDIUM EQUATION
• The Variational Formulation of the Fokker--Planck Equation
• Title not available (Why is that?)
• Generalized Iterative Scaling for Log-Linear Models
• Ricci curvature of Markov chains on metric spaces
• Ricci curvature of finite Markov chains via convexity of the entropy
• Fokker-Planck equations for a free energy functional or Markov process on a graph
• Geodesic convexity of the relative entropy in reversible Markov chains
• Information Theory and Statistics: A Tutorial
• Gradient flows of the entropy for finite Markov chains
• Ricci curvature of graphs
• Constrained steepest descent in the 2-Wasserstein metric
• Generalization of an inequality by Talagrand and links with the logarithmic Sobolev inequality
• Ricci curvature and eigenvalue estimate on locally finite graphs
• Entropic Ricci curvature bounds for discrete interacting systems
• Discrete Ricci curvature bounds for Bernoulli-Laplace and random transposition models
• Ollivier's Ricci curvature, local clustering and curvature-dimension inequalities on graphs
• Geometric analysis aspects of infinite semiplanar graphs with nonnegative curvature
• A curved Brunn-Minkowski inequality on the discrete hypercube, or: what is the Ricci curvature of the discrete hypercube?
• The Density Manifold and Configuration Space Quantization
• Ricci curvature bounds for weakly interacting Markov chains
• Earth mover's distances on discrete surfaces
• Information geometry and its applications
• Poincaré, modified logarithmic Sobolev and isoperimetric inequalities for Markov chains with non-negative Ricci curvature
• Natural gradient via optimal transport
• Information geometry
• Gradient flow formulation and longtime behaviour of a constrained Fokker-Planck equation
• Super Ricci flows for weighted graphs

Cited In (20)

• Diffusion hypercontractivity via generalized density manifold
• Wasserstein statistics in one-dimensional location scale models
• Wasserstein statistics in one-dimensional location-scale models
• Title not available (Why is that?)
• Optimal transport natural gradient for statistical manifolds with continuous sample space
• Wasserstein proximal of GANs
• Quantum statistical learning via quantum Wasserstein natural gradient
• Neural parametric Fokker-Planck equation
• Natural gradient for combined loss using wavelets
• Wasserstein information matrix
• Transport information geometry: Riemannian calculus on probability simplex
• Hessian transport gradient flows
• Wasserstein Riemannian geometry on statistical manifold
• Transport information Bregman divergences
• On some connections between Esscher's tilting, saddlepoint approximations, and optimal transportation: a statistical perspective
• Information geometry of Wasserstein statistics on shapes and affine deformations
• When optimal transport meets information geometry
• Hessian metric via transport information geometry
• Logarithmic divergences from optimal transport and Rényi geometry
• Natural gradient via optimal transport

This page was built for publication: Ricci curvature for parametric statistics via optimal transport

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2006199)