Applying a New Grid-Based Elitist-Reserving Strategy to EMO Archive Algorithms

Abstract

Grid-based measure is an often-used strategy by some MOEAs to maintain the diversity of the solution sets. The well known ε-MOEA, based on the ε-dominance concept, is essentially based on grid-strategy too. Though often gaining an appropriate tradeoff between the aspects of the performance, the ε-MOEA has its inherent vice and behaves unacceptably sometimes. That is, when the PF_true’s slope to one dimension changes a lot along the coordinate, the algorithm loses many extreme or representative individuals, that has obvious influence on the diversity of the solution sets. In order to solve this problem, a new δ-dominance concept and the suppositional optimum point concept are defined. Then we proposed a new grid-based elitist-reserving strategy and applied it in an EMO archive algorithm (δ-MOEA). The experimental results illustrated δ-MOEA’s good performance, which is much better especially for the diversity than NSGA-II and ε-MOEA.