Crisscross Harris Hawks Optimizer for Global Tasks and Feature Selection.

Area of Science:

• Computational Intelligence
• Metaheuristic Optimization
• Machine Learning

Background:

• Harris Hawks Optimizer (HHO) is an effective metaheuristic algorithm inspired by hawk hunting behaviors.
• Original HHO exhibits limitations in global search capability due to its levy distribution.
• Crisscross Optimization Algorithm (CSO) offers strategies to improve exploration and exploitation.

Purpose of the Study:

• To introduce a hybrid optimization algorithm, the Crisscross Harris Hawks Optimizer (CCHHO).
• To address the global search limitations of the standard Harris Hawks Optimizer.
• To enhance convergence speed and solution quality in optimization problems.

Main Methods:

• Integration of CSO's vertical and horizontal crossover strategies into HHO.
• Implementation of a competitive operator to accelerate convergence.
• Adaptive adjustment of exploitation and exploration using CSO's techniques.

Main Results:

• CCHHO demonstrated significantly improved performance over HHO, CSO, and other state-of-the-art algorithms.
• The proposed CCHHO achieved accelerated convergence and high-quality solutions across benchmark functions, engineering problems, and feature selection tasks.
• CCHHO's effectiveness was maintained regardless of problem dimensionality.

Conclusions:

• The Crisscross Harris Hawks Optimizer (CCHHO) effectively overcomes the limitations of the original Harris Hawks Optimizer.
• CCHHO offers a robust and efficient approach for various optimization challenges, including feature selection and engineering design.
• The hybrid strategy significantly enhances both the exploration and exploitation phases of the optimization process.