Adaptive Differential Evolution with Competent Leaders

Differential evolution (DE) has been proven to be a powerful population-based optimization algorithm, successfully used in various scientific and engineering fields. However, in DE, the search is guided by either a random vector or a local optimal vector. Inspired by the natural phenomenon of that good species usually contain good information, this paper propose a competent leaders guiding strategy (cLGS) for DE. In cLGS, the population is firstly divided into different clusters to construct new individuals as the competent leaders, and then the leaders compete with the top individuals of the population to guide the search. With the competent leaders, the good information of the population can be utilized effectively for guiding the search. In order to test the efficiency of the proposed strategy, cLGS is incorporated into JADE which is a very competitive DE variant. Twenty benchmark functions are chosen to experimental validated the proposed algorithm. Its high performance is confirmed by comparing with several DE variants.