Understanding and Accelerating Particle-Based Variational Inference

Abstract

Particle-based variational inference methods (ParVIs) have gained attention in the Bayesian inference literature, for their capacity to yield flexible and accurate approximations. We explore ParVIs from the perspective of Wasserstein gradient flows, and make both theoretical and practical contributions. We unify various finite-particle approximations that existing ParVIs use, and recognize that the approximation is essentially a compulsory smoothing treatment, in either of two equivalent forms. This novel understanding reveals the assumptions and relations of existing ParVIs, and also inspires new ParVIs. We propose an acceleration framework and a principled bandwidth-selection method for general ParVIs; these are based on the developed theory and leverage the geometry of the Wasserstein space. Experimental results show the improved convergence by the acceleration framework and enhanced sample accuracy by the bandwidth-selection method.

arXiv commentA typo for citation corrected

Cited by in corpus (24)

• On the geometry of Stein variational gradient descent
• Repulsive Deep Ensembles are Bayesian
• Particle-based Energetic Variational Inference
• Information Newton's flow: second-order optimization method in probability space
• Path Integral Sampler: a stochastic control approach for sampling
• Stein Self-Repulsive Dynamics: Benefits From Past Samples
• On Stein Variational Neural Network Ensembles
• Annealed Flow Transport Monte Carlo
• SVGD as a kernelized Wasserstein gradient flow of the chi-squared divergence
• Hessian-Free High-Resolution Nesterov Acceleration for Sampling
• De-randomizing MCMC dynamics with the diffusion Stein operator
• Probabilistic Iterative Methods for Linear Systems
• Frank-Wolfe Methods in Probability Space
• Variational Transport: A Convergent Particle-BasedAlgorithm for Distributional Optimization
• The equivalence between Stein variational gradient descent and black-box variational inference
• Self-Paced Deep Reinforcement Learning
• Stabilizing Training of Generative Adversarial Nets via Langevin Stein Variational Gradient Descent
• Accelerated Information Gradient flow
• Multilevel Stein variational gradient descent with applications to Bayesian inverse problems
• Relative Entropy Gradient Sampler for Unnormalized Distributions
• On the Generative Utility of Cyclic Conditionals
• Generative Particle Variational Inference via Estimation of Functional Gradients
• A Probabilistic Interpretation of Self-Paced Learning with Applications to Reinforcement Learning
• A Wasserstein Minimum Velocity Approach to Learning Unnormalized Models