Adaptation of genetic operators and parameters of a genetic algorithm based on the elite degree of an individual
Koichi Hatta, Shin’ichi Wakabayashi, Tetsushi Koide · Systems and Computers in Japan · 2000
Genetic algorithms (GAs) are known as a heuristic technique for solving large-scale combinatorial optimization problems having many constraints. Their performance depends heavily on the setting of the GA parameters that control execution, and on the types and probabilities of application of the genetic operators. In the context of optimal choice of the above factors, the authors previously proposed the elite degree as a measure of individuals' latent fitness; this measure was then employed to develop a new GA allowing adaptive selection among crossover options. In this study, the previous definition of elite degree is revised; using the new definition, a GA is developed to support adaptive adjustment of mutation probability. This allows faster attainment of better solutions. The effectiveness of the proposed method is confirmed by experiments. © 2000 Scripta Technica, Syst Comp Jpn, 32(1): 29–37, 2001