Enhanced grey wolf optimizer with hybrid strategies for efficient feature selection in high-dimensional data

Feature selection (FS) is a critical data preprocessing technique in the field of machine learning. It aims to identify the most relevant subset of features while simultaneously minimizing their number. The Grey Wolf Optimizer (GWO) is widely adopted for feature selection due to its high accuracy and low parameter requirements. However, the traditional GWO tends to fall into local optima when dealing with high-dimensional FS tasks. In this paper, a Multi-strategy Improved GWO (MIGWO) is proposed for solving high-dimensional feature selection problems: Firstly, the distribution of the initial grey wolf population is optimized based on the ReliefF algorithm for calculating feature importance to improve the convergence speed; secondly, the position updating equation is expanded by combining the weighting of fitness values and the competitive mechanism to improve the exploitation; finally, the hybrid GWO with differential evolution algorithm (DE) and Levy-flight strategy are used to advance the exploration. Experimental results on 10 high-dimensional datasets demonstrate that the MIGWO obtains a smaller feature subset while obtaining a higher classification accuracy compared with the current mainstream feature selection methods, which proves the effectiveness and superiority of the proposed algorithm in dealing with high-dimensional feature selection problems.