Gbest-guided covariance matrix adaptation evolution strategy for large scale global optimization

Abstract

The optimization of a large number of decision variables, so called large scale global optimization (LSGO) remains challenging for existing heuristics. Inspired by the concept of global best (gbest) guided strategy, this paper proposes a gbest-guided covariance matrix adaptation evolution strategy (GCMA-ES) where the gbest information is utilized in the search equation to guide the exploitation process. The GCMA-ES can take advantages from both the CMA-ES and the gbest-guided strategy. Its performance is demonstrated on the CEC 2010 LSGO benchmarks.