Firefly algorithm with multiple learning ability based on gender difference.

Area of Science:

• Computational Intelligence
• Optimization Algorithms

Background:

• The standard Firefly Algorithm (FA) faces challenges like search oscillation and limited convergence precision.
• Existing FA variants often struggle to effectively balance exploration and exploitation for complex optimization tasks.

Purpose of the Study:

• To introduce a novel Firefly Algorithm with Multiple Learning Ability based on Gender Difference (MLFA-GD).
• To address the limitations of traditional FA, specifically search oscillation and low convergence precision.

Main Methods:

• The MLFA-GD algorithm divides the population into male and female subgroups.
• It employs distinct learning strategies: males use a partial attraction model with an escape mechanism, while females are guided by the male centroid and global optimum.
• A random walk strategy is integrated to refine optimization accuracy.

Main Results:

• Experiments on 23 numerical functions and 30 CEC 2017 benchmark functions demonstrated superior performance.
• Comparisons with six FA variants and ten metaheuristic algorithms showed enhanced search capability and higher optimization precision.
• The MLFA-GD effectively balances exploration and exploitation.

Conclusions:

• The proposed MLFA-GD significantly improves upon existing Firefly Algorithm variants.
• The gender-based learning strategies and random walk enhance both exploration and exploitation capabilities.
• MLFA-GD offers a more precise and effective optimization solution for complex problems.