Adaptive Directional Mutation for an Adaptive Differential Evolution Algorithm

Differential Evolution (DE) has been successfully applied to various optimization problems. The performance of DE is affected by algorithm parameters, mutation strategies and so on. One of the most successful studies on controlling the parameters is JADE. In this study, we focus on mutation strategies and propose an adaptive mutation strategy to improve JADE. Some directional mutation strategies, which utilize directional difference vectors from bad individuals to good individuals, have been proposed to improve search speed. However, in order to avoid local solutions, the vectors in opposite direction may be necessary. The proper frequency of such vectors depends on the problem and the search process. We propose an adaptive mutation strategy that controls the frequency of directional and opposite directional vectors adaptively to realizes efficient and stable search. The advantage of JADE with the proposed method is shown by solving thirteen benchmark problems.