Controlling Hybrid Evolutionary Algorithms in Subset Selection for Multimodal Optimization

Multimodal functions play a central role in artificial intelligence and evolutionary algorithms. Still, there are several limitations when it comes to existing work on optimization of complex multimodal functions. In this paper, we study the optimization of such functions, in the subset selection problem setting, by carefully integrating different methods: evolutionary algorithms, stochastic local search, clustering, and feedback control. The goal is to carefully balance exploration and exploitation during hybrid evolutionary search by using feedback control to adapt population diversity via crowding. We empirically test our integrated method on complex synthetic combinatorial optimization problems, demonstrating promising results compared to previous work.