Explaining Variational Approximations

被引：237

作者：

Ormerod, J. T. ^{[1
]}

Wand, M. P. ^{[2
]}

机构：

[1] Univ Sydney, Sch Math & Stat, Sydney, NSW 2006, Australia

[2] Univ Wollongong, Ctr Stat & Survey Methodol, Sch Math & Appl Stat, Wollongong, NSW 2522, Australia

来源：

AMERICAN STATISTICIAN | 2010年 / 64卷 / 02期

基金：

澳大利亚研究理事会;

关键词：

Bayesian inference; Bayesian networks; Directed acyclic graphs; Generalized linear mixed models; Kullback-Leibler divergence; Linear mixed models; BAYESIAN MODEL SELECTION; INFERENCE;

D O I：

10.1198/tast.2010.09058

中图分类号：

O21 [概率论与数理统计]; C8 [统计学];

学科分类号：

020208 ; 070103 ; 0714 ;

摘要：

Variational approximations facilitate approximate inference for the parameters in complex statistical models and provide fast, deterministic alternatives to Monte Carlo methods. However, much of the contemporary literature on variational approximations is in Computer Science rather than Statistics, and uses terminology, notation, and examples from the former field. In this article we explain variational approximation in statistical terms. In particular, we illustrate the ideas of variational approximation using examples that are familiar to statisticians.

引用

页码：140 / 153

页数：14

共 36 条

[1] BAYESIAN-ANALYSIS OF BINARY AND POLYCHOTOMOUS RESPONSE DATA [J].

ALBERT, JH ;

CHIB, S .

JOURNAL OF THE AMERICAN STATISTICAL ASSOCIATION, 1993, 88 (422) :669-679

[2]

[Anonymous], THESIS U NEW S WALES

[3]

[Anonymous], 2006, Pattern recognition and machine learning

[4]

Archambeau C., 2007, Journal of Machine Learning Research Workshop and Conference Proceedings, V1, P1

[5]

Barber D, 1998, ADV NEUR IN, V10, P395

[6]

Boyd S.P, 2004, Convex optimization, DOI [DOI 10.1017/CBO9780511804441, 10.1017/CBO9780511804441]

[7] EXPLAINING THE GIBBS SAMPLER [J].

CASELLA, G ;

GEORGE, EI .

AMERICAN STATISTICIAN, 1992, 46 (03) :167-174

[8] Mean-field variational approximate Bayesian inference for latent variable models [J].

Consonni, Guido ;

Marin, Jean-Michel .

COMPUTATIONAL STATISTICS & DATA ANALYSIS, 2007, 52 (02) :790-798

[9] Variational Bayesian multinomial probit regression with gaussian process priors [J].

Girolami, Mark ;

Rogers, Simon .

NEURAL COMPUTATION, 2006, 18 (08) :1790-1817

[10] On the adequacy of variational lower bound functions for likelihood-based inference in Markovian models with missing values [J].

Hall, P ;

Humphreys, K ;

Titterington, DM .

JOURNAL OF THE ROYAL STATISTICAL SOCIETY SERIES B-STATISTICAL METHODOLOGY, 2002, 64 :549-564

← 1 2 3 4 →