Evolving ensemble of classifiers in low-dimensional spaces using multi-objective evolutionary approach


In this paper we discuss a new strategy to create ensemble of classifiers based on the multi objective evolutionary optimization. Instead of using feature selection technique which has been widely used in multi objective evolutionary approaches for ensemble generating, we have used a bagging-and-boosting-like strategy which also covers problems with lower dimensional feature spaces in which using feature selection technique may lead to ambiguous subspaces. After creating classifiers based on the amount of error created for each class, a multi-objective genetic algorithm has used to combine them to provide a set of powerful ensembles. Comprehensive experiments demonstrate the effectiveness of the proposed strategy.