Bayesian network structure learning based on cuckoo search algorithm

Bayesian network is a graphical model based on probabilities to represent and inference in uncertain conditions. In the field of Bayesian network, structure learning from data is an important challenge. One of the methods to learn structure of a Bayesian Networks is score and search method. In this paper, a new method for learning the structure of the Bayesian network is presented by using Cuckoo Optimization Algorithm. In accordance with the Cuckoo search algorithm, in the proposed algorithm, a population of directed acyclic graphs is created, which are equivalent to the cuckoos. Each graph has a score which indicates its fitness. The algorithm is repeated until it finds the best solution or an acceptable network structure. In each iteration, it searches for the directed acyclic graph with the best score. Based on the practical results obtained, the proposed algorithm has a better performance than the other algorithms and offers higher scores.