An Efficient Fitness Assignment Based on Dominating Tree
Chuan Shi, Zhongzhi Shi, Bin Wu · 2007
It has seen a surge of research activity on multiobjective optimization using evolutionary algorithms in recent years. The majority of these algorithms use fitness assignment based on Pareto dominance. The fitness assignment not only decides the algorithm's performance, but also is one of the main time-consuming components. This paper proposes an efficient fitness assignment based on dominating tree (DT). The dominating tree is a binary tree with the dominating information of individuals, which can represent three-valued relationship existing in Pareto dominance. We apply the dominating tree as an effective fitness assignment that can improve general multiobjective evolutionary algorithms. The simulation results also prove it.