Expectation Propagation for Infinite Mixtures

Thomas Minka and Zoubin Ghahramani
NIPS'03 Workshop on Nonparametric Bayesian Methods and Infinite Models

This note describes a method for approximate inference in infinite models that uses deterministic Expectation Propagation instead of Monte Carlo. For infinite Gaussian mixtures, the algorithm provides cluster parameter estimates, cluster memberships, and model evidence. Model parameters, such as the expected size of the mixture, can be efficiently tuned via EM with EP as the E-step. The same approach can apply other infinite models such as infinite HMMs.

Extended abstract PS . Extended abstract PDF . Talk PS . Talk PDF


Tom Minka
Last modified: Mon Apr 26 12:29:45 GMT 2004