* Ph.D. Thesis Summary:
The research aimed to enhance the Slime Mould Algorithm (SMA) to address limitations such as slow convergence and local optima. Two advanced variants were developed: the Smart Switching Slime Mould Algorithm (S2SMA) for large-scale problems, including face sketch recognition with deep learning models, and the Merged Slime Mould Algorithm (MSMA) for low-dimensional problems, integrating Adaptive Opposition SMA (AOSMA) and Vertical Smart Switching Rules (VSSR).
The proposed algorithms were evaluated on face sketch datasets, engineering design problems, and benchmark optimization tests. Results showed significant improvements in convergence speed, recognition accuracy, and adaptability. Key contributions include embedded rules for improved exploration-exploitation, integration with multiple deep learning-based face recognition models, and enhanced operational efficiency. Overall, the study provides robust advancements in metaheuristic optimization and deep learning-based face sketch recognition.