摘要：This paper presents a new hybrid approach for learning Bayesian networks (BNs) based on artificial bee colony algorithm and particle swarm optimization. Firstly, an unconstrained optimization problem is established, which can provide a smaller search space. Secondly, the definition and encoding of the basic mathematical elements of our algorithm are given, and the basic operations are designed, which provide guarantee of convergence. Thirdly, from a known original Bayesian network with probabilistic logic sampling, full samples for the training set and testing set are generated, and then the structure of Bayesian network is learned from complete training set by using our method. The simulation experimental results show that our method is effective.
