Learning Bayesian networks for clustering by means of constructive induction

The purpose of this paper is to present and evaluate a heuristic algorithm for learning Bayesian networks for clustering. Our approach is based upon i mproving the Naive-Bayes model by means of constructive induction. A key id ea in this approach is to treat expected data as real data. This allows us to complete the database and to take advantage of factorable closed forms f or the marginal likelihood. In order to get such an advantage, we search fo r parameter values using the EM algorithm or another alternative approach t hat we have developed: a hybridization of the Bound and Collapse method and the EM algorithm, which results in a method that exhibits a faster converg ence rate and a more effective behaviour than the EM algorithm. Also, we co nsider the possibility of interleaving runnings of these two methods after each structural change. We evaluate our approach on synthetic and real-worl d databases. (C) 1999 Elsevier Science B.V. All rights reserved.