Binary Fish Migration Optimization for Feature Selection Using Time-Varying Transfer Function

Fish migration optimization (FMO) is an intelligent optimization algorithm based on fish migration behavior. It has been proven superior performance in terms of convergence, accuracy, and speed. Binary fish migration optimization (BFMO) is the binary version of FMO, which is designed to solve generator set assembly problem. In this paper, time-varying binary fish migration optimization (T-v-BFMO) is proposed to balance the exploration and exploitation of the BFMO. Then, experiments are made on various phi(max) and phi(min). The exact value of phi(max) and phi(min) is selected by comparing the performance of various phi values on 23 standard benchmark functions. In the end, T-v-BFMO successfully implements feature selection on UCI datasets.