Dynamic Inertia Weight Binary Bat Algorithm with Neighborhood Search

Area of Science:

• Computer Science
• Artificial Intelligence
• Optimization Algorithms

Background:

• Binary Bat Algorithm (BBA) is a variant of the Bat Algorithm (BA) for binary optimization problems.
• BBA shows competitiveness but can suffer from premature convergence due to velocity update mechanisms inherited from BA.
• Existing heuristic algorithms may also face challenges in escaping local minima.

Purpose of the Study:

• To propose an Improved Binary Bat Algorithm (IBBA) to overcome the premature convergence issue in BBA.
• To enhance the global search capability and solution quality of binary optimization algorithms.
• To provide a more robust optimization technique for complex problems.

Main Methods:

• The study introduces modifications to the velocity update process within the BBA framework to create IBBA.
• Performance evaluation involved standard mathematical benchmark functions.
• The algorithm's efficacy was further tested on the zero-one knapsack problem.

Main Results:

• IBBA significantly outperformed the original BBA and Binary Particle Swarm Optimization (BPSO) on benchmark functions.
• Experiments on zero-one knapsack problems confirmed IBBA's enhanced ability to avoid local minima compared to other heuristic algorithms.
• Numerical results indicate improved convergence behavior and solution accuracy.

Conclusions:

• The proposed Improved Binary Bat Algorithm (IBBA) effectively mitigates premature convergence.
• IBBA presents a more robust and efficient optimization approach for binary problems.
• This enhanced algorithm offers a valuable alternative for tackling complex optimization tasks, including the knapsack problem.