A Reinforcement-based Mechanism to Select Features for Classifiers in Ensemble Systems

Abstract

Classifier ensemble are systems composed of a set of individual classifiers structured in a parallel way and a combination module, which is responsible for providing the final output of the system. One way of increasing diversity in classifier ensembles is to use feature selection methods in order to select subsets of attributes for the individual classifiers. In this paper, it is investigated the use of a simple reinforcementbased method, called ReinSel, in ensemble systems. This method is inserted into the filter approach of feature selection methods and it chooses only the attributes that are important only for a specific class through the use of a reinforcement procedure.