Venue: IEEE Symposium Series on Computational Intelligence (SSCI)
Year: 2019
Paper: https://ieeexplore.ieee.org/document/9002937
Abstract
In many real-world applications, multi-objective optimization problems may have more than one Pareto sets. The goal of multi-modal multi-objective optimization is to find all Pareto sets in the decision space. As a relatively new research area, difficulties in solving multi-modal multi-objective optimization problems have not been carefully analyzed in the literature. In this paper, first we point out that standard evolutionary multi-objective algorithms show difficulties when solving multimodal multi-objective optimization problems. Next, using the concept of genetic drift, we clearly explain the decrease of diversity in the decision space during the evolutionary process. Then, we report performance evaluation results of state-of-the-art evolutionary multi-modal multi-objective algorithms using scalable test problems.
Additional information
If the IEEE link doesn’t work for you, here is the Researchgate copy of the paper.