Covariance Matrix adaptation based on Opposition learning for multimodal optimization
Wei Li, Qingzheng Xu · 2019
Although CMA-ES is effective in locating a single global optimum, it cannot perform well on multi-optima problems. In order to improve the performance of CMA-ES, this paper proposed Covariance Matrix adaptation based on Opposition learning (CMA-OL). In CMA-OL, an improved dynamic peak identification method is introduced in CMA-OL to identify the various peaks dynamically. Opposition learning method is employed in CMA-OL to improve the probability of visiting unproductive regions of the search space. To verify the effectiveness of CMA-OL, numerical experiments are carried on six benchmark problems from CEC2013. The experimental results show that CMA-OL is competitive with respect to other compared algorithms for solving multi-optima problems.