Gender-Based Deep Learning Firefly Optimization Method for Test Data Generation

Area of Science:

• Computer Science
• Artificial Intelligence
• Software Engineering

Background:

• Software testing is crucial for quality assurance.
• Intelligent optimization algorithms aid automatic test data generation.
• The standard firefly algorithm has limitations in convergence and accuracy.

Purpose of the Study:

• To propose a novel firefly algorithm integrated with deep learning for structural test data generation.
• To enhance the convergence rate and accuracy of the firefly algorithm.
• To improve software quality assurance through effective test data generation.

Main Methods:

• Dividing the firefly population into male and female subgroups.
• Implementing a global search for male fireflies attracted to random female fireflies.
• Utilizing deep learning for a central firefly to guide local search for female fireflies.
• Incorporating chaos search near the best firefly in the final stage.

Main Results:

• The proposed algorithm demonstrates superior performance compared to standard methods.
• Achieved higher success coverage rates.
• Reduced coverage time.
• Increased the diversity of generated solutions.

Conclusions:

• The deep learning-enhanced firefly algorithm is effective for structural test data generation.
• This approach offers significant improvements in software testing efficiency and effectiveness.
• The method provides a promising direction for advanced software quality assurance techniques.