A Feature Selection Method using PSO-MI

Feature selection method is used for generating an optimal number of features to be used for a certain task like classification. Particle Swarm Optimization (PSO) is an algorithm influenced by the habit of bird flocking or fish schooling. The main goal of this paper is designing and evaluating a viable wrapper-based feature selection algorithm. Feature selection can also be viewed as a minimization problem for which PSO can be applied. In this paper, we attempt to introduce a PSO based feature selection method using mutual information (MI). Feature-class MI has been used to select a subset of features based on its relevancy. A wrapper-based method is used to find the productiveness of the method by evaluating with different classifiers in different datasets. The classification performances have been found promising when compared with classifications performed using normal classifiers and PSO method without using mutual information.