A Non-Asymptotic Analysis for Stein Variational Gradient Descent

Abstract

We study the Stein Variational Gradient Descent (SVGD) algorithm, which optimises a set of particles to approximate a target probability distribution $\pi \propto e^{-V}$ on ${\mathbb{R}}^d$. In the population limit, SVGD performs gradient descent in the space of probability distributions on the KL divergence with respect to $\pi$, where the gradient is smoothed through a kernel integral operator. In this paper, we provide a novel finite time analysis for the SVGD algorithm. We provide a descent lemma establishing that the algorithm decreases the objective at each iteration, and rates of convergence for the average Stein Fisher divergence (also referred to as Kernel Stein Discrepancy). We also provide a convergence result of the finite particle system corresponding to the practical implementation of SVGD to its population version.

arXiv commentAccepted to Neurips 2020

References in corpus (13)

• Rényi Divergence and Kullback-Leibler Divergence
• Neural Tangent Kernel: Convergence and Generalization in Neural Networks
• User-friendly guarantees for the Langevin Monte Carlo with inaccurate gradient
• Stein Variational Gradient Descent as Gradient Flow
• A Kernel Test of Goodness of Fit
• Measuring Sample Quality with Kernels
• Learning to Draw Samples: With Application to Amortized MLE for Generative Adversarial Learning
• A Stein variational Newton method
• On the geometry of Stein variational gradient descent
• Rapid Convergence of the Unadjusted Langevin Algorithm: Isoperimetry Suffices
• Learning to Draw Samples with Amortized Stein Variational Gradient Descent
• Stein Variational Gradient Descent as Moment Matching
• VAE Learning via Stein Variational Gradient Descent

Cited by in corpus (5)

• Repulsive Deep Ensembles are Bayesian
• Convergence of Stein Variational Gradient Descent under a Weaker Smoothness Condition
• Stein Variational Gaussian Processes
• Sampling with Mirrored Stein Operators
• Variational Transport: A Convergent Particle-BasedAlgorithm for Distributional Optimization