Nonparametric Clustering with Dirichlet Process Mixture Model

Abstract

Clustering is one of the most useful techniques in machine learning and data mining. In cluster analysis, model selection which is how to determine the number of clusters is an important issue. Unlike in supervised learning, there are no class labels and criteria to guide the search, so the model selection for clustering is always difficult. To tackle this problem, we present the concept of nonparametric clustering approach based on Dirichlet process mixture model (DPMM), and apply a Gibbs sampling technique to sample the posterior distribution. The proposed clustering algorithm follows the nonparametric Bayesian framework and can optimize the number of components and the parameters of the model. The experimental result of clustering shows that this Bayes model has attractive properties and robust performance, and the number of clusters can be determined automatically.