An adaptive configuration of differential evolution algorithms for big data


Recent years have witnessed the success of evolutionary algorithms, such as differential evolution, in solving many complex optimization problems. However, evolutionary algorithms face difficulties when solving high dimensional problems in conjunction with a huge amount of data. Due to the fact that no single differential evolution algorithm is the best for all types of test problems, in this paper, an adaptive configuration of differential evolution algorithms is proposed for solving the 2015 big data optimization competition problems. In it, different differential evolution variants are used. The proposed algorithm automatically determines the best variant, which is subsequently used to evolve a population of individuals till the end of evolutionary process. Additionally, a local search is applied to increase the exploitation capability of the proposed algorithm. The results of solving six single objective problems, and six multi-objective problems, show the superiority of the proposed algorithm to the baseline algorithm's results provided by the competition organizers.