Open Access
Translator Disclaimer
December 2010 Particle learning for general mixtures
Carlos M. Carvalho, Hedibert F. Lopes, Nicholas G. Polson, Matt A. Taddy
Bayesian Anal. 5(4): 709-740 (December 2010). DOI: 10.1214/10-BA525

Abstract

This paper develops particle learning (PL) methods for the estimation of general mixture models. The approach is distinguished from alternative particle filtering methods in two major ways. First, each iteration begins by resampling particles according to posterior predictive probability, leading to a more efficient set for propagation. Second, each particle tracks only the "essential state vector" thus leading to reduced dimensional inference. In addition, we describe how the approach will apply to more general mixture models of current interest in the literature; it is hoped that this will inspire a greater number of researchers to adopt sequential Monte Carlo methods for fitting their sophisticated mixture based models. Finally, we show that PL leads to straightforward tools for marginal likelihood calculation and posterior cluster allocation.

Citation

Download Citation

Carlos M. Carvalho. Hedibert F. Lopes. Nicholas G. Polson. Matt A. Taddy. "Particle learning for general mixtures." Bayesian Anal. 5 (4) 709 - 740, December 2010. https://doi.org/10.1214/10-BA525

Information

Published: December 2010
First available in Project Euclid: 19 June 2012

zbMATH: 1330.62348
MathSciNet: MR2740154
Digital Object Identifier: 10.1214/10-BA525

Keywords: Dirichlet process , Indian buffet process , Mixture models , nonparametric , particle filtering , probit stick-breaking

Rights: Copyright © 2010 International Society for Bayesian Analysis

JOURNAL ARTICLE
32 PAGES


SHARE
Vol.5 • No. 4 • December 2010
Back to Top