Compressed-Coding Particle Swarm Optimization for Large-Scale Feature Selection

Abstract

AbstractParticle swarm optimization (PSO) is a popular method for feature selection. However, when dealing with large-scale features, PSO faces the challenges of poor search performance and long running time. In addition, a suitable representation for particles to deal with the discrete binary optimization problem like the feature selection is still in great need. This paper proposes a PSO algorithm for large-scale feature selection problems named compressed-coding PSO (CCPSO). It uses the N-base encoding method for the representation of particles and designs a particle update mechanism based on the Hamming distance, which can be performed in the discrete space. It also proposes a local search strategy to dynamically shorten the length of particles, thus reducing the search space. The experimental results show that CCPSO performs well for large-scale feature selection problems. The solutions obtained by CCPSO contain small feature subsets and have an excellent performance in classification problems.KeywordsParticle swarm optimizationLarge-scale feature selectionCompressed codingHamming distance