A Diversity-Control-oriented Genetic Algorithm (DCGA) : Development and Initial Experimental Results (Special Issue on New System Software)

概要

論文の詳細を見る
In genetic algorithms, in order to attain the global optimum without getting stuck at a local optimum, appropriate diversity of the structures in the population needs to be maintained. I propose a new genetic algorithm called DCGA (Diversity-Control-oriented Genetic Algorithm) to attain this goal. In the DCGA, the structures of the population in the next generation are selected from the merged population of parents and their offspring on the basis of a selection probability, which is calculated by using a hamming distance betWeen a candidate structure and the structure with the best fitness value, and is larger for structures with larger hamming distances. The diversity of structures in the population can be externally controlled by adjusting the coefficients of the pfobability function so as to be in an appropriate condition according to the given problem. Within the range of my experiments, the DCGA showed a markedly superior performance to the simple GA and it seems to be a promising competitor of previously proposed algorithms.
一般社団法人情報処理学会の論文
1999-06-15