Hybrid Beta-Henry Optimization A Novel Feature Selection Approach | ||||
| IJCI. International Journal of Computers and Information | ||||
| Article 7, Volume 12, Issue 2, July 2025, Page 81-94 PDF (1.68 MB) | ||||
| Document Type: Original Article | ||||
| DOI: 10.21608/ijci.2025.331259.1182 | ||||
 View on SCiNiTO | ||||
| Authors | ||||
Shadia Elgazzar  ; Hatem Mohamed; Asmaa Haroun
| ||||
| Department of Information Systems, Faculty of Computer Science and Information Systems, Menoufiya University, Shebin El-Koom, Egypt | ||||
| Abstract | ||||
| Inspired by nature, the Henry Gas Solubility Optimization (HGSO) algorithm is a novel approach for solving global optimization problems by simulating Henry's Law of gas solubility. However, premature convergence and an imbalanced exploration-exploitation ratio remain significant challenges. HGSO's simple search strategy limits its ability to exploit optimal solutions effectively, hindering its performance on complex optimization problems. To address these limitations, we propose a beta hill climbing local search to enhance HGSO's performance. This novel beta operator improves upon traditional hill climbing by carefully balancing exploration and exploitation. By incorporating this operator, the resulting Enhanced HGSO (EHGSO) can more efficiently traverse the solution space and identify optimal solutions. We evaluated EHGSO on nine benchmark datasets and a real-world dataset. For the real-world dataset, we employed a Random Forest model, while for the benchmark datasets, we used a KNN model. EHGSO achieved an accuracy of 0.9467 on the real-world dataset and consistently outperformed other meta-heuristics like GOA, WOA, DA, GWO, and SSA on the benchmark datasets. These results demonstrate the superior optimization capabilities of EHGSO in tackling complex optimization problems. | ||||
| Keywords | ||||
| Henry Gas Solubility Optimization; Machine learning; Local Search; Optimization Algorithms | ||||
|
Statistics Article View: 113 PDF Download: 37 |
||||
