Ensemble based classifiers using dictionary learning

Dictionary learning is used in signal, image, audio and video processing applications to represent signals by a sparse set of atoms where sparse representations are managed for the problems of compression, denoising, feature extraction and data classification. In many machine learning applications, classifier ensembles are shown to be superior than their single classifier counterparts. In this paper, we propose to use dictionary learning as a base classifier in ensemble learning methods and introduce Random Subspace Dictionary Learning (RDL) and Bagging Dictionary Learning (BDL) algorithms by learning ensembles of dictionaries for each class using feature/instance subspaces. The experimental results show that the ensemble based dictionary learning methods outperform the single dictionary learning (DL), Support Vector Machines (SVM), and SVM based ensemble classifiers.