Cooperative coevolution using the Brain Storm Optimization Algorithm

The Brain Storm Optimization (BSO) algorithm is a powerful optimization approach that has been proposed in the past few years. A number of improvements have been previously proposed for BSO with successful application for real-world problems. In this paper, the implementation of a Cooperative Co-evolutionary BSO (CCBSO) algorithm that is based on the explicit space decomposition approach is investigated. The improved performance of CCBSO is illustrated based on experimental comparisons with other BSO variants on a library of 20 well-known classical benchmark functions.

I. Introduction

The concept of cooperative algorithms was proposed to have a more efficient search mechanism in comparison to the individual instances of the same algorithm. Cooperation is based on having multiple algorithm instances searching the search space (either serially or in parallel) and exchanging information useful information. This implies that the search space is decomposed between the different algorithms [1], [2]. The decomposition is either explicit by dividing the problem variables between the different cooperating instances or implicit by having the different instances search different parts of the search space due to different initialization, different parameters settings or both.