A GA-Based Approach for Parameter Learning of Discrete Dynamic Bayesian Networks

Learning dynamic Bayesian networks (DBNs) is one of the current research focuses. In this article a GA-based approach is proposed for DBNs parameters learning from fully and partially observed data. The validity of the novel approach has been demonstrated by a detailedly described example, and the experimental results show that the proposed GA-based approach performs more accurately than the traditional EM algorithm.