A hierarchical multi-leadership sine cosine algorithm to dissolving global optimization and data classification: The...

Area of Science:

• Computational Intelligence
• Optimization Algorithms
• Machine Learning Applications

Background:

• The Sine Cosine Algorithm (SCA) is a widely used optimizer for complex problems.
• Existing SCA versions suffer from insufficient population diversity and a poor balance between exploration and exploitation.
• Addressing these limitations is crucial for improving optimization performance in real-world applications.

Purpose of the Study:

• To introduce an improved Sine Cosine Algorithm, termed Hierarchical Multi-Leadership SCA (HMLSCA).
• To enhance population diversification and the exploration-exploitation balance in optimization.
• To validate the efficacy of HMLSCA across benchmark functions, medical data classification, and COVID-19 diagnosis.

Main Methods:

• Development of the Hierarchical Multi-Leadership Sine Cosine Algorithm (HMLSCA).
• Evaluation using 18 classical benchmark functions and 30 CEC 2017 test suites.
• Application to optimize Support Vector Machine (SVM) parameters and feature weighting for medical data classification.
• Deployment for COVID-19 diagnosis using a dedicated dataset.

Main Results:

• HMLSCA demonstrated superior performance compared to established metaheuristic algorithms on benchmark functions.
• The algorithm achieved the highest classification accuracy in medical data tasks, evidenced by a Friedman mean rank of 1.00.
• HMLSCA attained a 98% accuracy rate in diagnosing COVID-19 infections.

Conclusions:

• The proposed HMLSCA effectively addresses the shortcomings of the original SCA, offering improved population diversity and exploration-exploitation balance.
• HMLSCA exhibits promising efficiency and outperforms existing algorithms in various optimization and classification tasks.
• The algorithm's successful application in medical data classification and COVID-19 diagnosis highlights its practical utility and effectiveness.