Sequential MCMC for Bayesian Model Selection

Sequential MCMC for Bayesian Model Selection (1999)

Cached

Download Links

[www.cs.berkeley.edu]

by Christophe Andrieu , Nando de Freitas , Arnaud Doucet

Venue:	IEEE Higher Order Statistics Workshop, Ceasarea
Citations:	29 - 15 self

Versions

Version History

Metadata Version 3

User correction supplied by SystemCorrections

Datum	Value	Source
TITLE	Sequential MCMC for Bayesian Model Selection	user correction - Legacy Corrections
AUTHOR NAME	Christophe Andrieu	user correction - Legacy Corrections
AUTHOR AFFIL	Department of Engineering, University of Cambridge	user correction - Legacy Corrections
AUTHOR ADDR	Trumpington Street, CB2 1PZ Cambridge, UK	user correction - Legacy Corrections
AUTHOR NAME	Nando de Freitas	user correction - Legacy Corrections
AUTHOR AFFIL	Department of Engineering, University of Cambridge	user correction - Legacy Corrections
AUTHOR ADDR	Trumpington Street, CB2 1PZ Cambridge, UK	user correction - Legacy Corrections
AUTHOR NAME	Arnaud Doucet	user correction - Legacy Corrections
AUTHOR AFFIL	Department of Engineering, University of Cambridge	user correction - Legacy Corrections
AUTHOR ADDR	Trumpington Street, CB2 1PZ Cambridge, UK	user correction - Legacy Corrections
ABSTRACT	In this paper, we address the problem of sequential Bayesian model selection. This problem does not usually admit any closed-form analytical solution. We propose here an original sequential simulation-based method to solve the associated Bayesian computational problems. This method combines sequential importance sampling, a resampling procedure and reversible jump MCMC moves. We describe a generic algorithm and then apply it to the problem of sequential Bayesian model order estimation of autoregressive (AR) time series observed in additive noise. 1 Introduction Model selection is a fundamental data analysis task. It has many applications in various fields of science and engineering. Over the past two decades many of these problems have been addressed using information criteria such as AIC or MDL. The widespread use of these criteria is mainly due to their intrinsic simplicity. AIC and MDL are applied by evaluating two terms: a data term which requires the maximization of the likeliho...	user correction - Legacy Corrections
YEAR	1999	INFERENCE
VENUE	IEEE Higher Order Statistics Workshop, Ceasarea	INFERENCE
VENUE TYPE	CONFERENCE	INFERENCE
PAGES	130--134	INFERENCE
CITATIONS	16 found	ParsCit 1.0