Differential evolution algorithm with dichotomy-based parameter space compression

Differential evolution (DE) is a very simple, but effective technique for solving various optimization problems. However, the performance of DE remarkably relies on its control parameter settings, and enormous adaptive or self-adaptive mechanisms for DE have been proposed to improve the robustness of DE. In this paper, we put forward an enhanced parameter adaptation technique for DE, which exploits the previous successful experience to compress the parameter space by using the dichotomy (called DPADE, i.e., dichotomy-based parameter adaptation DE). In this way, the control parameters are able to approach the suitable values for the given problems. The proposed technique is integrated with three classic mutation operators and one state-of-the-art mutation operator. The experimental results on 59 problems derived from the CEC2014 benchmark set and CEC2017 benchmark set show that our proposed method is able to improve the performance of DE and it is more effective than other state-of-the-art parameter control techniques.