Infinite mixtures of trees

Finite mixtures of tree-structured distributions have been shown to be efficient and effective in modeling multivariate distributions. Using Dirichlet processes, we extend this approach to allow countably many treestructured mixture components. The resulting Bayesian framework allows us to deal with the problem of selecting the number of mixture components by computing the posterior distribution over the number of components and integrating out the components by Bayesian model averaging. We apply the proposed framework to identify the number and the properties of predominant precipitation patterns in historical archives of climate data.

• Type: Conference
• Year: 2007
• Where: ICML
• Authors: Sergey Kirshner, Padhraic Smyth
• Keywords: Bayesian Model Averaging, ICML 2007, Machine Learning, Predominant Precipitation Patterns, Treestructured Mixture Components