On Tuning Group Sizes in the Random Adaptive Grouping Algorithm for Large-scale Global Optimization Problems

Abstract

Large-scale global optimization (LSGO) is known as one of the most challenging problem for evolutionary algorithms (EA). The most advanced algorithms for LSGO are based on cooperative coevolution with problem decomposition using grouping methods. In our previous studies, we have proposed a novel random adaptive grouping algorithm (RAG) that combines the ideas of random dynamic grouping and learning dynamic grouping. We have demonstrated that an approach based on the DECC and the RAG outperforms some state-of-the-art LSGO algorithms on the IEE CEC LSGO benchmarks. In this study, we have investigated the problem of tuning group sizes within the decomposition stage in details. We have evaluated the performance of the DECC-RAG algorithm with LSGO 2010 and 2013 benchmarks. The results of numerical experiments are presented and discussed. The results demonstrates how the performance of the RAG depends on the group sizing for each type of LSGO problems.