A multi-strategy improved hunger games search algorithm

This paper proposes a Multi-strategy Improved Hunger Games Search (MHGS) algorithm to address the inherent limitations of the original HGS algorithm, including imbalanced exploration-exploitation capabilities, insufficient population diversity, and premature convergence. The main contributions feature four synergistic innovation mechanisms: (1) A phased position update framework dynamically coordinates global exploration and local exploitation through three distinct search phases; (2) An enhanced reproduction operator mimics biological reproductive patterns to maintain population diversity; (3) An adaptive boundary handling system redirects out-of-bounds individuals to promising regions, improving search efficiency; (4) An elite dynamic oppositional learning strategy with self-adjusting coefficients enhances local optima avoidance. The proposed mechanisms demonstrate synergistic effects: the phased update coordinates macro/micro-search patterns, while the reproduction operator and boundary handling jointly maintain solution diversity, complemented by oppositional learning's perturbation effects. Extensive evaluations on 23 benchmark functions, CEC2017 test suite, and two engineering designs reveal MHGS's superior performance, achieving 23.7% average accuracy improvement over seven state-of-the-art algorithms (Wilcoxon rank-sum test p < 0.05). Furthermore, the binary variant BMHGS_V3 with sigmoid transformation attains 92.3% average classification accuracy on ten UCI datasets for feature selection. The proposed algorithm establishes a novel framework for complex optimization, demonstrating both theoretical significance and practical value in computational intelligence.