A Hierarchical Sorting Swarm Optimizer for Large-Scale Optimization

Large-scale optimization is a challenging problem because it involves a large number of decision variables. In this paper, a simple but effective method, called hierarchical sorting swarm optimizer (HSSO), is proposed for large-scale optimization. As a variant of representative particle swarm optimizer (PSO), HSSO first sorts the initial particles according to their fitness values, and then partitions the sorted particles into two groups, namely, the good group corresponding to better fitness values, and the bad group with worse fitness values. The bad group is then updated by learning from the good one. After that, we take the good group as a new swarm and conduct the sorting and learning procedures. The aforementioned operations are repeated several times until only one particle left to form a hierarchical structure. In the experiments, HSSO is applied to optimize 39 benchmark test functions. The comparative results with several existing algorithms demonstrate that, despite its simplicity, HSSO shows improved performance in terms of both exploration and exploitation.