The aim of this paper is to provide a perspective for evolutionary computing in the context of research into adaptive intelligent systems. The author proposes a problem class hierarchy, by which problems can be classified into four classes. Traditional AI techniques provide a powerful basis for several classes of problems that can be solved by linear methods or linear approximation. However, solutions to highly nonlinear problems and nonequilibrium environment problems have not yet been established. The author argues that evolutionary computing and emergent computing provide effective problem solving methods for broad range of problems in all classes.
抄録全体を表示