Bayesian Inference for Dynamic Models with Dirichlet Process Mixtures

Author

Caron, Francois ; Davy, Manuel ; Doucet, Arnaud ; Duflos, Emmanuel ; Vanheeghe, Philippe

Author_Institution

Ecole Centrale de Lille, LAGIS, Villeneuve d´´Ascq

fYear

2006

fDate

10-13 July 2006

Firstpage

1

Lastpage

8

Abstract

Using Kalman techniques, it is possible to perform optimal estimation in linear Gaussian state-space models. We address here the case where the noise probability density functions are of unknown functional form. A flexible Bayesian nonparametric noise model based on mixture of Dirichlet processes is introduced. Efficient Markov chain Monte Carlo and sequential Monte Carlo methods are then developed to perform optimal estimation in such contexts

Keywords

Bayes methods; Gaussian noise; Kalman filters; Markov processes; Monte Carlo methods; inference mechanisms; probability; state-space methods; Bayesian inference; Dirichlet process mixtures; Kalman techniques; Markov chain Monte Carlo methods; linear Gaussian state-space models; noise probability density functions; Bayesian methods; Computer science; Context modeling; Deconvolution; Gaussian noise; Monte Carlo methods; Particle filters; Probability density function; State estimation; Statistics; Bayesian nonparametrics; Dirichlet Process Mixture; Monte Carlo Markov Chain; Particle filter; Rao-Blackwellisation;

fLanguage

English

Publisher

ieee

Conference_Titel

Information Fusion, 2006 9th International Conference on

Conference_Location

Florence

Print_ISBN

1-4244-0953-5

Electronic_ISBN

0-9721844-6-5

Type

conf

DOI

10.1109/ICIF.2006.301580

Filename

4085866