APSM-jS']jSO: A novel jS']jSO variant with an adaptive parameter selection mechanism and a new external archive updating mechanism

A novel jSO variant named APSM-jSO is proposed in this study by making simple and effective modifications to improve its performance. There are three main differences between APSM-jSO and jSO. First, a novel adaptive selection mechanism (APSM) for selecting entries from the historical memory is designed to fully utilize the better entries in the historical memory. Subsequently, the first in-first out (FIFO) method is utilized to update the external archive for maintaining the population diversity and avoiding overuse of the external archive. Finally, a new mutation strategy adopting rank-based selective pressure (RSP) is used to enhance the exploitation of APSM-jSO. APSM-jSO is evaluated using the IEEE CEC 2018 test suite in comparison with five state-of-the-art DE-based variants (ELSHADE-SPACMA, EB-LSHADE, LSHADE-RSP, mL-SHADE, and MadDE) and five winners of the IEEE CEC competitions (AGSK, APGSK-IMODE, EBOwithCMAR, HSES, and IMODE). The results demonstrate that APSM-jSO outperforms jSO and five DE-based algorithms, is superior to four top winners of the IEEE CEC competitions (AGSK, APGSK-IMODE, HSES, and IMODE), and is not inferior to the top method of the IEEE CEC 2017 competition (EBOwithCMAR). The conclusion is that APSM-jSO effectively further enhances the overall performance of jSO, and is an excellent jSO variant. The MATLAB source code of APSM-jSO can bedownloaded from https://github.com/Yintong-Li/APSM-jSO.