Novel binary encoding water cycle algorithm for solving Bayesian network structures learning problem

Constructing Bayesian network structures from data is a computationally hard task. One important method to learn Bayesian network structures uses the meta-heuristic algorithms. In this paper, a novel binary encoding water cycle algorithm is proposed for the first time to address the Bayesian network structures learning problem. In this study, the sea, rivers and streams correspond to the candidate Bayesian network structures. Since it is a discrete problem to find an optimal structure, the logic operators have been used to calculate the positions of the individuals. Meanwhile, to balance the exploitation and exploration abilities of the algorithm, the ways how rivers and streams flow to the sea and the evaporation process have been designed with the new strategies. Experiments on well-known benchmark networks demonstrate that the proposed algorithm is capable of identifying the optimal or near-optimal structures. In the comparison to the use of the other algorithms, our method performs well and turns out to have the better solution quality.