Improved binary particle swarm optimization for feature selection with new initialization and search space reduction strategies

An-Da Li, Bing Xue, Mengjie Zhang

Abstract: Highlights•An improved binary PSO algorithm is proposed for feature selection.•A new initialization strategy using the feature weighting results is proposed.•A bits masking strategy for dynamically reducing the search space is proposed.•The genetic operations are adopted to alleviate the premature convergence problem.•The proposed algorithm is effective and efficient for feature selection.