A Novel Cooperative Co-Evolutionary Framework for Large-Scale Overlapping Problems

Abstract

Large-scale overlapping problems are a type of challenging problems because of the special structure of the shared variables. In order to solve the large-scale overlapping problems effectively, the paper proposes a novel cooperative co-evolutionary framework for large-scale overlapping problems called the cooperative co-evolution algorithm with subgroup merging and resource allocation for overlapping problems (CCMRO). In CCMRO, a contribution-based subcomponent merging strategy is proposed to deal with the decomposition of the large-scale overlapping problem and reduce the interaction of subgroups. In addition, a new computational resource allocation strategy (CCRO) is designed to maintain a balance between the full allocation of computational resources and the stability of search speed. Finally, the effectiveness of the proposed CCMRO is verified by the large-scale overlapping benchmark functions.