Particle methods for maximum likelihood estimation in latent variable models

Standard methods for maximum likelihood parameter estimation in latent variable models rely on the Expectation-Maximization algorithm and its Monte Carlo variants. Our approach is different and motivated by similar considerations to simulated annealing; that is we build a sequence of artificial dist...

Full description

Saved in:
Bibliographic Details
Published inStatistics and computing Vol. 18; no. 1; pp. 47 - 57
Main Authors Johansen, Adam M., Doucet, Arnaud, Davy, Manuel
Format Journal Article
LanguageEnglish
Published Boston Springer US 01.03.2008
Subjects
Artificial Intelligence
Mathematics and Statistics
Probability and Statistics in Computer Science
Statistical Theory and Methods
Statistics
Statistics and Computing/Statistics Programs
Sequential Monte Carlo
Latent variable models
Maximum likelihood
Markov chain Monte Carlo
Simulated annealing
Online AccessGet full text

Cover

More Information
Summary:Standard methods for maximum likelihood parameter estimation in latent variable models rely on the Expectation-Maximization algorithm and its Monte Carlo variants. Our approach is different and motivated by similar considerations to simulated annealing; that is we build a sequence of artificial distributions whose support concentrates itself on the set of maximum likelihood estimates. We sample from these distributions using a sequential Monte Carlo approach. We demonstrate state-of-the-art performance for several applications of the proposed approach.
ISSN:0960-3174
1573-1375
DOI:10.1007/s11222-007-9037-8