We propose a parallel genetic algorithm with sexual selection including global selection and adaptive division of subpopulations. This method can search the diverse directions of evolution in the phenotype space. Global selection which is applied to all females in global population changes the number of females in each subpopulation, and adjusts each weight of the search. Simulation of the proposed method shows subpopulations with the high average fitness values are increased and divided into two, while subpopulations with the low average fitness values are extinguished. As a result, the proposed method increases the efficiency of the search.
展开▼
