Scaling Up the Greedy Equivalence Search Algorithm by Constraining the Search Space of Equivalence Classes

Abstract

Greedy Equivalence Search (GES) is nowadays the state of the art algorithm for learning Bayesian networks (BNs) from complete data. However, from a practical point of view, this algorithm may not be fast enough to work in high dimensionality domains. This paper proposes some variants of GES aimed to increase its efficiency. Under faithfulness assumption, the modified algorithms preserve the same theoretical properties as the original one, that is, they recover a perfect map of the target distribution in the large sample limit. Moreover, experimental results confirm that, although they carry out much less computations, BNs learnt by those algorithms have the same quality as those learnt by GES.