Swarm Intelligence-Based Feature Selection: An Improved Binary Grey Wolf Optimization Method

Feature selection can effectively reduce the number of features and improve the accuracy of classification, so that reducing the computational burden and improving the performance of machine learning. In this paper, we propose an improved binary grey wolf optimization (IBGWO) algorithm for a wrapper-based feature selection method. Aiming at the shortcomings of grey wolf optimization (GWO) for feature selection, we first propose a enhanced opposition-based learning (E-OBL) initialization method to enhance the performance of initial solutions. Second, a local search strategy is introduced to balance the exploitation and exploration abilities of the IBGWO. Finally, a novel update mechanism is proposed for improving the population diversity and exploration capability of the algorithm. Simulations are conducted by using 16 well-known datasets, and the results show that the proposed method outperforms other benchmark algorithms on 12 datasets, and the introduced improved factors are suitable and effective.