Differential Evolution (DE) is simple but powerful algorithm for optimization problems. Offspring are generated by the crossover between parents and mutant vectors. There are two methods for selecting the base vector in the mutant vector, best and rand. It is difficult to select the appropriate strategy according to the problems. In this paper, we propose a new species-based DE method for searching a global optimum. Our speciation methods utilize Beta-skeleton graph and K-means clustering. The search space is divided into multiple sub-spaces by the speciation, and respective species search for various solutions. As a result, a global optimum can be acquired more efficiently. We applied our proposed methods to some function optimization problems and showed their effectiveness.

2011 IEEE SMC Hiroshima Chapter若手研究会, 2011年7月9日, 広島市立大学