An Enhanced Grey Wolf Optimizer with Opposition, Mutation, and Local Search Strategy for Feature Selection

Feature selection is a pivotal process in machine learning and data analysis, aimed at improving model performance by identifying and utilizing the most relevant features while eliminating redundant and irrelevant ones. The Grey Wolf Optimizer (GWO), a metaheuristic algorithm based on the social hierarchy and hunting behavior of grey wolves, has proven to be an effective method for feature selection. However, it is also easy to get trapped in the local optima. Therefore, this study presents an enhanced Grey Wolf Optimizer with Opposition, Mutation, and Local Search Strategy (GWO-OMLSS) to boost exploration and exploitation of GWO in order to avoid the algorithm falling into local optima. The proposed method was validated using ten becnhmark data sets from the UCI machine learning repository and compared with state-of-the-art metaheuristic algorithm-based feature selections, such as GWO, PSO, ESSA, and CS. The Experimental results demonstrate that GWO-OMLSS outperforms the standard of GWO and other metaheuristic algorithms on all data sets except Ionosphere and Wine. The proposed method achieves the best an accuracy of $\mathbf{9 9. 7 \%}$ and $\mathbf{9 9. 8 \%}$ on the Exactly and M-of-n data sets, respectively, while also significantly reducing the number of features. It indicates that the proposed method not only obtains high accuracy, but also reduces the dimensionality of features effectively. It means that the proposed method can be considered as a promising algorithm for feature selection.