An improved Bayesian structural EM algorithm for learning Bayesian networks for clustering

The application of the Bayesian Structural EM algorithm to learn Bayesian networks for clustering implies a search over the space of Bayesian network structures alternating between two steps: an optimization of the Bayesian network parameters (usually by means of the EM algorithm) and a structural search for model selection. In this paper, we propose to perform the optimization of the Bayesian network parameters using an alternative approach to the EM algorithm: the BC+EM method. We provide experimental results to show that our proposal results in a more effective and eecient version of the Bayesian Structural EM algorithm for learning Bayesian networks for clustering.