An Enhanced Slime Mould Algorithm for Function optimization

Abstract

This study focuses on enhancement of one of the recently published metaheuristic algorithms known as slime mould algorithm (SMA). The slime mould algorithm has been shown to be a good competitive approach in the field of metaheuristics, however, it still suffers from poor exploitative behaviour, thus, suffers from slow convergence and lacks providing potentially better solutions which eventually requires an improvement. Considering the latter fact, this study attempts to further enhance the capability of the original version of slime mould algorithm so that it can be utilized for optimization problems as an even better approach. Therefore, Nelder-Mead (NM) simplex search method was utilized as an aiding structure to enhance the slime mould algorithm in terms of local search, as well. The constructed hybrid approach (SMA-NM) utilizes the slime mould algorithm for diversification and Nelder-Mead method for intensification which consequently enhances the algorithm due to better refinement of balance between exploration and exploitation stages. To assess the capability of the proposed approach, unimodal and multimodal benchmark functions were used. The performance of the proposed hybrid algorithm was tested against those test functions, in terms of exploitation, exploration, statistical significance and ranking. by comparing it with the grey wolf optimization, arithmetic optimization, and the original version of slime mould algorithms. The performed analyses have shown the proposed approach to be a greater competitive approach to deal with optimization problems.