Diversity-based JADE Algorithm

Abstract

In adaptive DE, the parameter p is used to balance the algorithm exploration and exploitation ability. Selection of suitable p in the evolution process is an ongoing challenge for adaptive DE. Population diversity is an effective measure showing the distribution of individuals in the search space. In this paper, use of diversity as a search strategy selection criterion for calculating p in adaptive DE is proposed. The proposed algorithm normalizes the calculated diversity to [0, 1] and calculates p according to the normalized diversity using an exponential function. Higher diversity yields higher p, providing the proposal with higher exploration ability, and vice versa. The accuracy of the proposed algorithm was experimentally compared to those of five other advanced DE variants using benchmark functions from IEEE Congress on Evolutionary Computation 2017. The results show that the proposed algorithm achieved the best performance among all variants for most benchmark functions.