Abstract:

Memetic algorithm is a combination of global search based on populations and local search based on individuals. It has a high global search ability and is used successfully in structure learning of Bayesian network. The principle of the particle swarm optimization algorithm is incorporated into the proposed basic genetic algorithm operating operator. By means of the ergodicity and randomizity of the chaos algorithm and a higher convergence speed of the cloud-based adaptive algorithm, a local search using a cloud-based chaotic mutation is proposed, which can avoid the local optimum and find out the best network structure. The experiment results reveal that this algorithm can be effectively used for BN structure learning.