Improving CIMDE with An Mixed Selection Method
Dan Chen, Xia Dahai, Xiong Caiquan, Gu Wei · 2020
Differential evolution(DE) is a popular evolution algorithm. How to select individuals for mutation operators is a very critical problem. Many researchers proposed improved mutation operators by selecting more outstanding individuals to produce mutated vectors. Collective information-powered mutation operator based differential evolution(CIMDE) proposed an improved collective information-powered mutation operator that uses multiple outstanding individuals as the heuristic information to release the selection pressure. But we observed that the mutation operator will be easily trapped into stagnation. So an improved mixed mutation operator named “current or rand-to-ci_ m best/1” is proposed to help individuals to departure from stagnation. Experiments show that this operator can improve the performance of the algorithm on CEC2013 bentmark functions.