ANALYSIS OF THE BEHAVIOR OF GENETIC ALGORITHMS WHEN LEARNING BAYESIANNETWORK STRUCTURE FROM DATA

In the last few years Bayesian networks have become a popular way of m odelling probabilistic relationships among a set of variables for a gi ven domain. For large domains, though, the construction of Bayesian ne tworks is a hard task and the number of possible structures and the nu mber of parameters for those structures can be huge. Trying to solve t his, some researchers have studied how this construction can be automa ted. This work analyzes the behaviour of genetic algorithms when perfo rming such automation. It is shown that the different ways in which ge netic algorithms can tackle the problem influence the results. (C) 199 7 Elsevier Science B.V.