Bayesian networks on dirichlet distributed vectors

Wray Buntine*, Lan Du, Petteri Nurmi

*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference proceeding contributionpeer-review

2 Citations (Scopus)

Abstract

Exact Bayesian network inference exists for Gaussian and multinomial distributions. For other kinds of distributions, approximations or restrictions on the kind of inference done are needed. In this paper we present generalized networks of Dirichlet distributions, and show how, using the two-parameter Poisson-Dirichlet distribution and Gibbs sampling, one can do approximate inference over them. This involves integrating out the probability vectors but leaving auxiliary discrete count vectors in their place. We illustrate the technique by extending standard topic models to "structured" documents, where the document structure is given by a Bayesian network of Dirichlets.

Original languageEnglish
Title of host publicationProceedings of the 5th European Workshop on Probabilistic Graphical Models, PGM 2010
Pages33-40
Number of pages8
Publication statusPublished - 2010
Event5th European Workshop on Probabilistic Graphical Models, PGM 2010 - Helsinki, Finland
Duration: 13 Sep 201015 Sep 2010

Other

Other5th European Workshop on Probabilistic Graphical Models, PGM 2010
CountryFinland
CityHelsinki
Period13/09/1015/09/10

Fingerprint Dive into the research topics of 'Bayesian networks on dirichlet distributed vectors'. Together they form a unique fingerprint.

Cite this