Bayesian Approach to Mixture Models for Discrimination

摘要

This paper develops a Bayesian mixture model approach to discrimination. The specific problem considered is the classification of mobile targets, from Inverse Synthetic Aperture Radar images. However, the algorithm developed is relevant to the generic classification problem. We model the data measurements from each target as a mixture distribution. A Bayesian formalism is adopted, and we obtain posterior distributions for the parameters of our mixture models. The distributions obtained are too complicated for direct analytical use in a classifier, so a Markov chain Monte Carlo (MCMC) algorithm is used to provide samples from the distributions. These samples are then used to make classifications of future data.