A multi-optimizer cooperative coevolution method for large scale optimization

Cooperative coevolution framework is an effective strategy to deal with large scale optimization problems. However, most studies on cooperative coevolution framework utilize the same optimizer for all subcomponents, which may not be effective enough. In this paper, we propose a novel multi-optimizer cooperative coevolution method for large scale optimization problems which randomly chooses an optimization strategy for each subcomponent independently. Four extensively used differential evolution algorithms are utilized as candidate optimizers. Two of them have good exploration properties while the other two have good exploitation properties. Experimental results utilizing differential grouping algorithm as decomposition strategy of the cooperative coevolution framework show that this multi-optimizer CC method performs better on most of the CEC'2010 large-scale global optimization (LSGO) benchmark functions than each single-optimizer CC framework where all of the subcomponents use the same optimizer. What is more, experimental results also show that this multi-optimizer CC method is suitable for not only fixed decomposition strategy (DG, XDG, and GDG) but also dynamic decomposition strategy (Delta Grouping).