A Novel Genetic Algorithm for Subspace Based Subclasssifier Selection
Fei Wang, Ming Yang · 2009
Ensemble learning constitutes one of the most popular directions in machine learning and data mining currently. And in ensemble learning, feature subspace selection and corresponding classifier ensemble for classification becomes the principal topic, in which base classifiers(also called subclassifiers) are generated by different subspaces. However, very little work has been done for effectively selecting the subclassifiers induced by different subspaces. In this paper, we introduce a novel Genetic Algorithm for subspace ensemble based subclassifier selection, that is, the newly developed algorithm attempts to select significant and relevant subclassifiers using genetic algorithm for improving the classification performance of ensemble. The experimental results show that the algorithm of this paper has better or comparable performance than those obtained by the well-known ensemble methods such as Bagging, AdaBoost and Random Subspace. Of course, how to determine the number of subclassifiers and the parameters used in genetic algorithm is our ongoing work.