Multiobjective Fuzzy Competitive Swarm Optimization for High-Dimensional Feature Selection

Feature selection (FS) is a fundamental technique in machine learning and data mining that aims to choose the most relevant and important features from a high-dimensional feature space. This process can enhance the performance and generalization ability of the classification model. However, classifying high-dimensional datasets presents challenges including high computational cost and stagnation in local optima. Evolutionary algorithms (EAs) have been widely applied in FS to mitigate these issues given their global search capabilities. In this study, we propose a multiobjective fuzzy competitive swarm optimization (MOFCSO) algorithm for FS on high-dimensional data. First, a fuzzy logic-based approach is proposed to classify the competitive particles, enabling better exploration of the search space. Then, a self-learning mechanism for failed particles is introduced to further enhance the global search. To demonstrate the proposed algorithm, we contrast it with multiple progressive algorithms. The experimental outcomes indicate the proposed algorithm is a valid method for choosing features in data with high dimensions.