A self-adaptive binary differential evolution algorithm for large scale binary optimization problems

This study proposes a new self-adaptive binary variant of a differential evolution algorithm, based on measure of dissimilarity and named SabDE. It uses an adaptive mechanism for selecting how new trial solutions are generated, and a chaotic process for adapting parameter values. SabDE is compared against a number of existing state of the art algorithms, on a set of benchmark problems including high dimensional knapsack problems with up to 10,000 dimensions as well as on the 15 learning based problems of the Congress on Evolutionary Computation (CEC 2015). Experimental results reveal that the proposed algorithm performs competitively and in some cases is superior to the existing algorithms.