Akademik Veri Yönetim Sistemi
Applied Soft Computing, cilt.201, 2026 (SCI-Expanded, Scopus)
The Artificial Bee Colony (ABC) algorithm is a widely used swarm intelligence method due to its simple structure and strong exploitation capability. However, it may experience performance loss in complex and multi-modal optimization problems due to its limited exploration capability and tendency toward early convergence. In this study, an optimization algorithm called Momentum-Guided Search and Starfish Exploration-based Artificial Bee Colony (MSE-ABC) is proposed to overcome these limitations. The proposed method preserves the observer and scout bee phases employed in the classical ABC algorithm while enhancing its global search capability with trigonometric exploration operators inspired by starfish optimization algorithm (SFOA). Furthermore, a momentum-based steering mechanism and dynamic exponential step size control are integrated to accelerate convergence during the search process and adaptively maintain the exploration-exploitation balance. The performance of the MSE-ABC algorithm was comprehensively evaluated through CEC 2022 benchmark tests, a five different real-world engineering design optimization problems, and deep learning hyperparameter optimization experiments conducted on the CIFAR-10 datasets. Experimental results demonstrate that the proposed method achieves superior or competitive performance when compared with eight state-of-the-art metaheuristic algorithms. The statistical significance of the obtained results was verified using the Wilcoxon signed-rank test and the Friedman test. The findings reveal that the MSE-ABC algorithm offers an effective and reliable optimization approach for complex and multi-modal optimization problems.