Geometrical Insights for Implicit Generative Modeling

DOI10.1007/978-3-319-99492-5_11zbMATH Open1518.68290arXiv1712.07822OpenAlexW2962750481MaRDI QIDQ6162298FDOQ6162298

Maxime Oquab, Léon Bottou, David Lopez-Paz, Author name not available (Why is that?)

Publication date: 28 June 2023

Published in: Braverman Readings in Machine Learning. Key Ideas from Inception to Current State (Search for Journal in Brave)

Abstract: Learning algorithms for implicit generative models can optimize a variety of criteria that measure how the data distribution differs from the implicit model distribution, including the Wasserstein distance, the Energy distance, and the Maximum Mean Discrepancy criterion. A careful look at the geometries induced by these distances on the space of probability measures reveals interesting differences. In particular, we can establish surprising approximate global convergence guarantees for the $1$-Wasserstein distance,even when the parametric generator has a nonconvex parametrization.

Full work available at URL: https://arxiv.org/abs/1712.07822

Recommendations

• A geometric view of optimal transportation and generative model
• On the capacity of deep generative networks for approximating distributions
• Optimal transport for generative models
• Multi-projection of unequal dimension optimal transport theory for generative adversary networks
• Some theoretical properties of GANs

Mathematics Subject Classification ID

Learning and adaptive systems in artificial intelligence (68T05)

Cites Work

• Equivalence of distance-based and RKHS-based statistics in hypothesis testing
• Approximation Theorems of Mathematical Statistics
• Theory of Reproducing Kernels
• Title not available (Why is that?)
• On the empirical estimation of integral probability metrics
• Integral Probability Metrics and Their Generating Classes of Functions
• Universality, Characteristic Kernels and RKHS Embedding of Measures
• Title not available (Why is that?)
• On the Asymptotic Distribution of Differentiable Statistical Functions
• Optimal Transport
• A course in metric geometry
• A characterization of distributions by mean values of statistics and certain probabilistic metrics
• Energy statistics: a class of statistics based on distances
• Theoretical foundations of the potential function method in pattern recognition learning
• Envelope Theorems for Arbitrary Choice Sets
• On the rate of convergence in Wasserstein distance of the empirical measure
• Constructive quantization: approximation by empirical measures
• On the optimal estimation of probability measures in weak and strong topologies
• Estimating Divergence Functionals and the Likelihood Ratio by Convex Risk Minimization
• Stochastic approximation with two time scales
• The Methods of Distances in the Theory of Probability and Statistics
• Complexity of random smooth functions on the high-dimensional sphere
• Convergence rate of linear two-time-scale stochastic approximation.
• The Distribution of Distance in a Hypersphere
• Gluing lemmas and Skorohod representations

Cited In (6)

• Processing simple geometric attributes with autoencoders
• Spurious Valleys in Two-layer Neural Network Optimization Landscapes
• Radon-Sobolev variational auto-encoders
• Title not available (Why is that?)
• Characterization of Gromov-type geodesics
• The ultrametric Gromov-Wasserstein distance

This page was built for publication: Geometrical Insights for Implicit Generative Modeling

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6162298)