Binary Invasive Weed Optimization Algorithm Approaches for Binary Optimization

Invasive Weed Optimization (IWO) Algorithm which is one of the population-based optimization algorithms has been recently developed by inspired from weed colonization. In a simple yet powerful optimization algorithm called the IWO algorithm, it is aimed to imitate the stability, adaptability and randomness of weed weeds. For the optimization problems with binary structured solution space, the basic IWO algorithm should be modified because its basic version is proposed for solving continuous optimization problems. In this study, three different adapted versions of IWO, IWObin1, IWObin2 and IWObin3 for short, are proposed for binary optimization. In the proposed methods to solve binary optimization problems, despite the fact that artificial weeds in the algorithm works on the continuous solution space, each weed position is converted to binary values, before the objective function is evaluated. In the first approach, search space is binarized by utilizing mod 2 process. In the second approach, sigmoid function is used to transform continuous space into binary search space. As for the last binary approach of IWO algorithm, this approach is carried out by means of tanh function for converting to binary values. The accuracy and performance of the proposed approaches have been examined on well-known 12 benchmark instances of uncapacitated facility location problem. The results obtained by IWObin1, IWObin2 and IWObin3 are compared each other by employing well-known small, medium and large sized twelve instances of UFLPs. The performance of the proposed approaches is also analyzed and compared in terms of convergence speed and running time (CPU time). The experimental results and comparisons show that proposed algorithm is an alternative and simple binary optimization method in terms of solution quality and robustness.