Cahiers du CEREMADE

Unité Mixte de Recherche du C.N.R.S. N°7534
 
Abstract : In this review, we first detail the inferencial difficulties related with the multimodal features of posterior distributions associated with mixtures of distributions, including the label-switching phenomenon. We then propose a pivotal quantity based on the MCMC output that can be used to construct a sensible posterior expectation. The second part of the review describes the various MCMC algorithms used for posterior simulation and points out their limitations, then introduces the population Monte Carlo alternative. We conclude with a short overview of extensions and applications of mixture models.
 
 
BAYESIAN MODELLING AND INFERENCE ON MIXTURES OF DISTRIBUTIONS
MARIN Jean-Michel, MENGERSEN K.L., ROBERT Christian P.
2004-12
09-02-2004
 
Université de PARIS - DAUPHINE
Place du Maréchal de Lattre De Tassigny - 75775 PARIS CEDEX 16 - FRANCE
Téléphone : +33 (0)1 44-05-49-23 - fax : +33 (0)1 44-05-45-99