DELVAUX - AN ENVIRONMENT THAT LEARNS BAYESIAN RULE-SETS WITH GENETIC ALGORITHMS

The paper describes an inductive learning environment called DELVAUX for classification tasks that learns PROSPECTOR-style, Bayesian rules from sets of examples. A genetic algorithm approach is employed in which a population consists of rule-sets that generate offspring through the exchange of rules. A bucket brigade algorithm for Bayesian rules, called reward punishment mechanism is introduced, which evaluates the performance of Bayesian rules within a rule-set. Moreover, we explore the use of multi-rule-set decision-making strategies to improve the learning performance of DELVAUX. Finally, empirical results that evaluate the learning performance of our environment are presented.