The use of version space controlled genetic algorithms to solve the Boole problem

Abstract

It is demonstrated that the VGA (version space guided genetic algorithm) is a particular instantiation of a more general class of systems, termed autonomous learning elements (ALEs). The basic components of an ALEs are discussed. The Boole problem posed by S. W. Wilson (1987) is introduced, and its expression in terms of the VGA framework is discussed. The details of the VGA system are given followed by a discussion of results. In particular, the performances of the VGA on two versions of the Boole problem are described and compared with those of classifier systems and decision trees.<>