Discriminative methods are used for increasing pattern recognition and classification accuracy. These methods can be used as discriminative transformations applied to features or they can be used as discriminative learning algorithms for the classifiers. Most of discriminative feature transformation measures don’t consider the classification method errors and information. In this paper, we propose a feature transformation method for support vector machine to consider both features discrimination and classification error.
To this end, we use Multi-Objective Particle Swarm Optimization (Multi-PSO), where we consider two mentioned criteria as objectives in Multi-PSO fitness function. Experimental results on UCI dataset show that the proposed Multi-PSO based feature transformation method outperform other conventional methods of feature transformation when it is used as a preprocessing step for SVM.