Adaptive Particle Swarm Optimizer for Feature Selection

The combinatorial nature of the Feature Selection problem has made the use of heuristic methods indispensable even for moderate dataset dimensions. Recently, several optimization paradigms emerged as attractive alternatives to classic heuristic based approaches. In this paper, we propose a new an adapted Particle Swarm Optimization for the exploration of the feature selection problem search space. In spite of the combinatorial nature of the feature selection problem, the investigated approach is based on the original PSO formulation and integrates wrapper-filter methods within uniform framework. Empirical study compares and discusses the effectiveness of the devised methods on a set of featured benchmarks.