A Gap-Based Memetic Differential Evolution (GaMeDE) Applied to Multi-modal Optimisation - Using Multi-objective Optimization Concepts

This paper presents a method that took second place in the GECCO 2020 Competition on Niching Methods for Multimodal Optimization. The method draws concepts from combinatorial multi-objective optimization, but also adds new mechanisms specific for continuous spaces and multi-modal aspects of the problem. GAP Selection operator is used to keep a high diversity of the population. A clustering mechanism identifies promising areas of the space, that are later optimized with a local search algorithm. The comparison between the top methods of the competition is presented. The document is concluded by the discussion on various insightson the problem instances and the methods, gained during the research.