Research on hybrid strategy Particle Swarm Optimization algorithm and its applications.

Area of Science:

• Computational Intelligence
• Optimization Algorithms
• Machine Learning

Background:

• Real-world optimization problems are increasingly complex and high-dimensional.
• Traditional Particle Swarm Optimization (PSO) struggles with local optima and slow convergence.
• Advanced algorithms are needed to address these limitations.

Purpose of the Study:

• To introduce a novel Hybrid Strategy Particle Swarm Optimization (HSPSO) algorithm.
• To enhance both global and local search capabilities for complex optimization tasks.
• To evaluate HSPSO's effectiveness in benchmark functions and a real-world feature selection problem.

Main Methods:

• Integration of adaptive weight adjustment, reverse learning, Cauchy mutation, and Hook-Jeeves strategy.
• Evaluation using CEC-2005 and CEC-2014 benchmark functions.
• Application to feature selection for the UCI Arrhythmia dataset.

Main Results:

• HSPSO demonstrated superior performance compared to standard PSO, DAIW-PSO, HBF-PSO, BOA, ACO, and FA.
• Achieved optimal results in best fitness, average fitness, and stability across benchmark tests.
• HSPSO yielded a high-accuracy classification model for the Arrhythmia dataset, outperforming traditional methods.

Conclusions:

• HSPSO is an effective and robust algorithm for complex optimization problems.
• The hybrid strategy significantly improves search capabilities and convergence.
• HSPSO offers a promising approach for feature selection in machine learning applications.