Learning Bayesian networks from incomplete data using evolutionary algorithms
James W. Myers, Kathryn Blackmond Laskey, Kenneth DeJong · 1999
This paper describes an evolutionary algorithm approach to learning Bayesian networks from incomplete data. This problem is characterized by a huge solution space with a highly multimodal landscape. State-of-the-art approaches all involve using deterministic approaches such as the expectation-maximization algorithm. These approaches are guaranteed to find local maxima, but do not explore the landscape for other modes. Our approach evolves the structure of the network and the missing data. We use a factorial design to choose a good set of parameters for selection, crossover, and mutation. We show that our algorithm produces accurate results for a classification problem with missing data.