Bayesian network structure learning based on restricted particle swarm optimization

The Bayesian network structure learning is one of the main research technologies in the field of data mining and knowledge discovery, while the search space of the network structure is relatively bigger, some proposed algorithms have some defects that the convergent speed is slow and the accuracy is poor. A kind of information theory combining particle swarm optimization algorithm is put forward, which uses mutual information to limit particle initialization, and makes the particle swarm optimization algorithm converge in a relatively short period of time, then an ASIA network is applied as the simulation model and the proposed algorithm is compared with K2 algorithm. Experimental results show that the proposed algorithm can rapidly and accurately get Bayesian network structures.