A Stein variational Newton method

Abstract

Stein variational gradient descent (SVGD) was recently proposed as a general purpose nonparametric variational inference algorithm [Liu & Wang, NIPS 2016]: it minimizes the Kullback-Leibler divergence between the target distribution and its approximation by implementing a form of functional gradient descent on a reproducing kernel Hilbert space. In this paper, we accelerate and generalize the SVGD algorithm by including second-order information, thereby approximating a Newton-like iteration in function space. We also show how second-order information can lead to more effective choices of kernel. We observe significant computational gains over the original SVGD algorithm in multiple test cases.

arXiv comment18 pages, 7 figures

References in corpus (10)

• Variational Inference with Normalizing Flows
• Dimension-independent likelihood-informed MCMC
• Likelihood-informed dimension reduction for nonlinear inverse problems
• An introduction to sampling via measure transport
• Stein Variational Gradient Descent as Gradient Flow
• Stein Variational Policy Gradient
• Stein Variational Adaptive Importance Sampling
• Riemannian Stein Variational Gradient Descent for Bayesian Inference
• Vprop: Variational Inference using RMSprop
• A general spline representation for nonparametric and semiparametric density estimates using diffeomorphisms

Cited by in corpus (20)

• Stein Variational Gradient Descent With Matrix-Valued Kernels
• A stochastic version of Stein Variational Gradient Descent for efficient sampling
• A Non-Asymptotic Analysis for Stein Variational Gradient Descent
• Bayesian inference of heterogeneous epidemic models: Application to COVID-19 spread accounting for long-term care facilities
• Deep composition of tensor-trains using squared inverse Rosenblatt transports
• Projected Stein Variational Gradient Descent
• Stein variational gradient descent with local approximations
• Particle-based Energetic Variational Inference
• Minimum Stein Discrepancy Estimators
• Greedy inference with structure-exploiting lazy maps
• Stein variational gradient descent on infinite-dimensional space and applications to statistical inverse problems
• Information Newton's flow: second-order optimization method in probability space
• Stein Variational Online Changepoint Detection with Applications to Hawkes Processes and Neural Networks
• SVGD as a kernelized Wasserstein gradient flow of the chi-squared divergence
• Handling the Positive-Definite Constraint in the Bayesian Learning Rule
• Sampling with Mirrored Stein Operators
• Stein variational reduced basis Bayesian inversion
• Bayesian Full-waveform Inversion with Realistic Priors
• Multilevel Stein variational gradient descent with applications to Bayesian inverse problems
• Neural Variational Gradient Descent