Publication: Enhanced slime mould algorithms for face sketch recognition and engineering problems.
| datacite.subject.fos | oecd::Engineering and technology::Electrical engineering, Electronic engineering, Information engineering::Electrical and electronic engineering | |
| dc.contributor.author | Ahmad Alhashash, Khaled Mohammad | |
| dc.date.accessioned | 2026-11-10T08:14:56Z | |
| dc.date.available | 2026-11-10T08:14:56Z | |
| dc.date.issued | 2025-03 | |
| dc.description.abstract | The Slime Mould Algorithm (SMA), inspired by the behaviour of Physarum polycephalum, offers a unique approach to solving optimization challenges. Despite its promising features, existing variants encounter limitations such as slow convergence and local optima traps. This thesis addresses these shortcomings by introducing two enhanced SMA variants: the Smart Switching Slime Mould Algorithm (S2SMA) and the Merged Slime Mould Algorithm (MSMA). S2SMA is designed to tackle large-scale problems, including the face sketch recognition problem, by embedding several search operations and control rules that augment the exploration exploitation balance, prevent slow convergence, and avoid getting stuck in local optima traps. In evaluating S2SMA's effectiveness, two public datasets, XM2VTS (295 sketch-photo pairs) and CUFSF (1194 sketch-photo pairs), were employed. Moreover, the MSMA resolves limitations encountered by S2SMA in dealing with low dimensional problems by combining S2SMA with the Adaptive Opposition Slime Mould Algorithm (AOSMA) and incorporating Vertical Smart Switching Rules (VSSR). VSSR's ability to dynamically switch algorithms based on problem characteristics significantly enhances adaptability and operational efficiency. The versatility of MSMA is demonstrated by investigating its applicability to seven engineering design problems: a cantilever beam problem, a welded beam design problem, a pressure vessel problem, a compression coil spring design problem, a multiple disc clutch brake problem, a speed reducer problem, and a gear train design problem. The performance of two new SMA variants is compared against existing SMA variants and other optimization algorithms. S2SMA's effectiveness is assessed on two face sketch databases, XM2VTS and CUFSF, as well as on CEC's 2010 large scale benchmark. Numerical results show that the enhanced optimizer achieves higher fitness values and improved face sketch recognition accuracy. Statistical analysis reveals S2SMA's significant outperformance, showing a rapid convergence curve compared to other optimization techniques. MSMA's performance is also evaluated using the basic CEC 2005 benchmark and ten real-world applications. The Wilcoxon rank-sum test validates the effectiveness of the proposed approach. Numerical findings highlight the superior fitness value achieved by the proposed strategy, while statistical results indicate MSMA's superior performance and rapid convergence. | |
| dc.identifier.uri | https://erepo.usm.my/handle/123456789/25354 | |
| dc.language.iso | en | |
| dc.title | Enhanced slime mould algorithms for face sketch recognition and engineering problems. | |
| dc.type | Resource Types::text::thesis::doctoral thesis | |
| dspace.entity.type | Publication | |
| oairecerif.author.affiliation | Universiti Sains Malaysia |