Particle Swarm Intelligence as Feature Selector in Ensemble Systems

Ensemble systems are composed of a set of individual classifiers, organized in a parallel way, that receive the input patterns and send their output to a combination method, which is responsible for providing the final output of the system. The use of feature selection methods in ensemble systems has been shown to be efficient, since it reduces the dimensionality while increases the diversity among the individual classifiers of these systems. This problem can be considered as a search problem, in which it aims to find different subsets of attributes which provide accurate and diverse ensemble systems. This paper presents the use of particle swarm optimization to select attributes for an ensemble system. This is achieved by using this technique to simultaneously maximize the individual diversity of the base classifiers and the group diversity of an ensemble system. In order to evaluate the possible solutions obtained by this technique, two filter-based evaluation criteria will be used. Filter-based criteria were chosen because they are independent of the learning algorithm and have a low computational cost.