A Multi-Level Environment for the Improvement of the Multi-Swarm Cooperative PSO Algorithm

In recent years, we assist to growing up interest of PSO algorithm in engineering applications and especially in machine learning. However, the PSO algorithm stills suffer from some weakness as a global search method harming its use in these fields. Many authors have introduced a lot of strategies to improve this algorithm, where the use of multiple swarms seems more promoting. Even if this strategy has significantly improved the performance of the PSO algorithm, it stills not yet enough. In this paper, we introduce a particular environment based on a hierarchical society coordinating the exploration and exploitation mechanisms to enhance the multi-swarm approach. Subjected to well-known optimization benchmark functions, the introduced strategy has proved its efficiency and effectiveness among the most popular versions to improve the potential of the PSO algorithm as a global optimization technique.