Improved learning of Bayesian networks

The search space of Bayesian Network struc­ tures is usually defined as Acyclic Directed Graphs (DAGs) and the search is done by lo­ cal transformations of DAGs. But the space of Bayesian Networks is ordered with respect to inclusion and it is natural to consider that a good search policy should take this into ac­ count. The first attempt to do this (Chick­ ering 1996) was using equivalence classes of DAGs instead of DAGs itself. This approach produces better results but it is significantly slower. We present a compromise between these two approaches. It uses DAGs to search the space in such a way that the ordering by inclusion is taken into account. This is achieved by repetitive usage of local moves within each equivalence class of DAGs. We show that this new approach produces bet­ ter results than the original DAGs approach without substantial change in time complex­ ity. We present empirical results, within the framework of heuristic search and Markov Chain Monte Carlo, provided through the Alarm dataset.