This paper analyzes the evolving behavior of the population in an extended genetic algorithm (GA), termed the GA with neutral mutations. The uniqueness of the GA is in extending the framework of the standard GAs by adopting the structured string representation and new genetic operations which bring neutral genetic changes in most cases. The evolving behavior on the trap function is compared with that of the standard GA for making clear why the GA with neutral mutations shows attractive results for deceptive problems.
展开▼