Slice Sampling Mixture Models

Kalli, Maria and Griffin, Jim E. and Walker, Stephen G. (2008) Slice Sampling Mixture Models. Centre for Health Services Studies, 23 pp. (Full text available)

PDF
Download (294kB)
[img]
Preview

Abstract

We propose a more efficient version of the slice sampler for Dirichlet process mixture models described by Walker (2007). This sampler allows the fitting of infinite mixture models with a wide–range of prior specification. To illustrate this flexiblity we develop a new nonparametric prior for mixture models by normalizing an infinite sequence of independent positive random variables and show how the slice sampler can be applied to make inference in this model. Two submodels are studied in detail. The first one assumes that the positive random variables are Gamma distributed and the second assumes that they are inverse–Gaussian distributed. Both priors have two hyperparameters and we consider their effect on the prior distribution of the number of occupied clusters in a sample. Extensive computational comparisons with alternative ”conditional” simulation techniques for mixture models using the standard Dirichlet process prior and our new prior are made. The properties of the new prior are illustrated on a density estimation problem.

Item Type: Research report (external)
Uncontrolled keywords: Dirichlet process; Markov chain Monte Carlo; Mixture model; Normalized Weights; Slice sampler
Subjects: Q Science > QA Mathematics (inc Computing science)
Divisions: Faculties > Social Sciences > School of Social Policy Sociology and Social Research > Centre for Health Services Studies
Depositing User: Tony Rees
Date Deposited: 07 Sep 2010 14:02
Last Modified: 25 Apr 2014 14:38
Resource URI: http://kar.kent.ac.uk/id/eprint/24721 (The current URI for this page, for reference purposes)
ORCiD (Kalli, Maria):
ORCiD (Griffin, Jim E.):
ORCiD (Walker, Stephen G.):
  • Depositors only (login required):

Downloads

Downloads per month over past year