Convergence properties of a general algorithm for calculating variational Bayesian estimates for a normal mixture model

被引：82
作者：
Wang, Bo [1 ]
Titterington, D. M. [2 ]
机构：
[1] Newcastle Univ, Sch Math & Stat, Newcastle Upon Tyne, Tyne & Wear, England
[2] Univ Glasgow, Dept Stat, Glasgow, Lanark, Scotland
来源：
BAYESIAN ANALYSIS | 2006年 / 1卷 / 03期
关键词：
Mixture model; Variational Bayes; Local convergence; Laplace approximation;
D O I：
10.1214/06-BA121
中图分类号：
O1 [数学];
学科分类号：
0701 ; 070101 ;
摘要：
In this paper we propose a generalised iterative algorithm for calculating variational Bayesian estimates for a normal mixture model and investigate its convergence properties. It is shown theoretically that the variational Bayesian estimator converges locally to the maximum likelihood estimator at the rate of O (1/n) in the large sample limit.
引用
收藏
页码：625 / 649
页数：25
相关论文
共 24 条
  • [1] [Anonymous], 2005, P 10 INT WORKSH ART
  • [2] Attias H, 1999, UNCERTAINTY IN ARTIFICIAL INTELLIGENCE, PROCEEDINGS, P21
  • [3] Attias H, 2000, ADV NEUR IN, V12, P209
  • [4] Beal M.J., 2003, Variational algorithms for approximate Bayesian inference
  • [5] Bhatia R., 2013, MATRIX ANAL
  • [6] Computational and inferential difficulties with mixture posterior distributions.
    Celeux, G
    Hurn, M
    Robert, CP
    [J]. JOURNAL OF THE AMERICAN STATISTICAL ASSOCIATION, 2000, 95 (451) : 957 - 970
  • [7] Corduneanu C. Bishop, 2001, Artif. Intell. Statist., V18, P27
  • [8] Duda R. O., 1973, PATTERN CLASSIFICATI
  • [9] Evans MJ., 2000, APPROXIMATING INTEGR
  • [10] Ghahramani Z, 2001, ADV NEUR IN, V13, P507
123