Regularized Stein Variational Gradient Flow

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: He, Ye, Balasubramanian, Krishnakumar, Sriperumbudur, Bharath K., Lu, Jianfeng
Format: Preprint
Published: 2022
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866916239589244928
author He, Ye
Balasubramanian, Krishnakumar
Sriperumbudur, Bharath K.
Lu, Jianfeng
author_facet He, Ye
Balasubramanian, Krishnakumar
Sriperumbudur, Bharath K.
Lu, Jianfeng
contents The Stein Variational Gradient Descent (SVGD) algorithm is a deterministic particle method for sampling. However, a mean-field analysis reveals that the gradient flow corresponding to the SVGD algorithm (i.e., the Stein Variational Gradient Flow) only provides a constant-order approximation to the Wasserstein Gradient Flow corresponding to the KL-divergence minimization. In this work, we propose the Regularized Stein Variational Gradient Flow, which interpolates between the Stein Variational Gradient Flow and the Wasserstein Gradient Flow. We establish various theoretical properties of the Regularized Stein Variational Gradient Flow (and its time-discretization) including convergence to equilibrium, existence and uniqueness of weak solutions, and stability of the solutions. We provide preliminary numerical evidence of the improved performance offered by the regularization.
format Preprint
id arxiv_https___arxiv_org_abs_2211_07861
institution arXiv
publishDate 2022
record_format arxiv
spellingShingle Regularized Stein Variational Gradient Flow
He, Ye
Balasubramanian, Krishnakumar
Sriperumbudur, Bharath K.
Lu, Jianfeng
Machine Learning
Numerical Analysis
Analysis of PDEs
Statistics Theory
Computation
The Stein Variational Gradient Descent (SVGD) algorithm is a deterministic particle method for sampling. However, a mean-field analysis reveals that the gradient flow corresponding to the SVGD algorithm (i.e., the Stein Variational Gradient Flow) only provides a constant-order approximation to the Wasserstein Gradient Flow corresponding to the KL-divergence minimization. In this work, we propose the Regularized Stein Variational Gradient Flow, which interpolates between the Stein Variational Gradient Flow and the Wasserstein Gradient Flow. We establish various theoretical properties of the Regularized Stein Variational Gradient Flow (and its time-discretization) including convergence to equilibrium, existence and uniqueness of weak solutions, and stability of the solutions. We provide preliminary numerical evidence of the improved performance offered by the regularization.
title Regularized Stein Variational Gradient Flow
topic Machine Learning
Numerical Analysis
Analysis of PDEs
Statistics Theory
Computation
url https://arxiv.org/abs/2211.07861