基于随机梯度 Fisher 评分的贝叶斯后验抽样

Jun, 2012

基于随机梯度 Fisher 评分的贝叶斯后验抽样

Bayesian Posterior Sampling via Stochastic Gradient Fisher Scoring

Sungjin Ahn, Anoop Korattikara, Max Welling

TL;DR本文介绍了一种基于Langevin方程和带随机梯度的算法来近似采样贝叶斯后验分布的方法，并通过引入贝叶斯中心极限定理扩展了该算法，使其在高混合比率时能够从后验的正态近似中采样，在慢混合比率时则模拟SGLD的行为，作为优化初始阶段的高效优化器。

Abstract

In this paper we address the following question: Can we approximately sample from a bayesian posterior distribution if we are only allowed to touch a small mini-batch of data-items for every sample we generate?. An algorithm based on the →