The Binary Equilibrium Optimization Algorithm with Sigmoid Transfer Functions

The binary version of bio-inspired approaches is used frequently in modern digital world. We therefore propose a new kind of binary version of the equilibrium optimization algorithm which was just raised before. Causing the universal usage for Sigmoid function, we carried out the simulation experiments on the values of β parameter for Sigmoid function. Results show that the value of β should be large in the interval of 2 to 5 for most unimodal and multimodal, and even bottom flat-like benchmark functions. Unfortunately, some of the multimodal and bottom flat-like benchmark functions need smaller value of β. Even more, some of them such as Salomon function lack the discipline for β parameter. Conclusions could be made that specific experiment should be carried out for given problems before real application.