New Binary Reptile Search Algorithms for Binary Optimization Problems

Area of Science:

• Computational Intelligence
• Optimization Algorithms
• Computer Science

Background:

• Binarizing continuous metaheuristics is crucial for solving NP-hard binary optimization problems.
• Challenges include transforming continuous populations to binary solutions while maintaining search effectiveness.

Purpose of the Study:

• To evaluate the performance of the Reptile Search Algorithm (RSA) on binary optimization problems.
• To investigate a two-step binarization process using transfer functions and binarization rules.

Main Methods:

• Applied the RSA to the Set Covering Problem and 0-1 Knapsack Problem.
• Implemented a two-step binarization strategy combining transfer functions and binarization rules.
• Conducted comparative experiments with Particle Swarm Optimization and Grey Wolf Optimizer.

Main Results:

• The RSA demonstrated competitive performance on both benchmark problems, showcasing flexibility and adaptability.
• The Z4 transfer function consistently improved performance across all tested algorithms.
• Effectiveness of binarization strategies significantly impacts continuous metaheuristics' performance in binary domains.

Conclusions:

• The RSA is a viable continuous metaheuristic for binary optimization tasks when appropriately binarized.
• The Z4 transfer function shows promise as a robust component for binary optimization.
• Careful selection of binarization strategies is essential for optimizing continuous metaheuristics for discrete problems.