Almost sure convergence of Titterington´s recursive estimator for mixture models

Author

Wang, Shaojun ; Zhao, Yunxin

Author_Institution

Dept. of Comput. Sci., Waterloo Univ., Ont., Canada

fYear

2002

fDate

2002

Firstpage

11

Abstract

Titterington (see Journal of Royal Statistical Society, B, vol.46, p.257-67, 1984) proposed a recursive parameter estimation algorithm for finite mixture models. However, due to the well known problem of singularities and multiple maximum, minimum and saddle points that are possible on the likelihood surfaces, convergence analysis has seldom been made in the past years. In this paper, under mild conditions, we show the global convergence of Titterington´s recursive estimator and its MAP variant for mixture models of the full regular exponential family.

Keywords

convergence; maximum likelihood estimation; recursive estimation; stochastic processes; MAP variant; Titterington recursive estimator; almost sure convergence; finite mixture models; full regular exponential family; global convergence; likelihood surfaces; mild conditions; recursive parameter estimation; saddle points; singularities; stochastic approximation; Bayesian methods; Computer science; Convergence; H infinity control; Maximum likelihood estimation; Parameter estimation; Recursive estimation; Stochastic processes; Taylor series;

fLanguage

English

Publisher

ieee

Conference_Titel

Information Theory, 2002. Proceedings. 2002 IEEE International Symposium on

Print_ISBN

0-7803-7501-7

Type

conf

DOI

10.1109/ISIT.2002.1023283

Filename

1023283