Cuckoo search with dual-subpopulation and information-sharing strategy
Jun Xi, Li Ming Zheng · International Journal of Computing Science and Mathematics · 2021
Cuckoo search (CS) algorithm is simple and powerful in dealing with the global optimisation problem. However, how to strike a good balance between exploration and exploitation in CS is still an open question. The paper proposes a modified CS with dual-subpopulation and information-sharing strategy (DSIS_CS). In DSIS_CS, the population is divided into two subpopulations which are assigned different update tasks. Then, random solutions are selected from the dissimilar subpopulations in order to avoid the results from easily falling into the local optima. In addition, the DSIS strategy can be incorporated into other state-of-the-art CS variants to improve their optimisation performance. Extensive experiments on 28 functions chosen from CEC 2013 have been carried out. The results suggest that the DSIS strategy helps both the CS and its variants to achieve a better trade-off between exploration and exploitation.