Bat Algorithm with Different Initialization Approaches for Numerical Optimization

In the field of applied engineering, function optimization has been extensively applied for the purpose of finding an optimal solution for the given problem. BAT Algorithm (BA) is population-based stochastic algorithm utilized to determine the both discrete and continuous kind of problems. The use of robust pattern for population initialization may lead to enhance the performance of optimization algorithms which results in prevention of premature convergence. In this paper, we implemented the BA with the probability distributions based on initialization and introduced the four new methods of BA Initialization such as BA initialized with Beta distribution B-BAT, an Exponential distribution E-BAT, Gamma distribution G-BAT and finally the Weibull distribution W-BAT. The empirical results conclude that our proposed techniques are more robust in term of fast convergence rate.