Active Two Phase Collaborative Representation Classifier for Image Categorization

Abstract

In recent times, the Sparse Representation Classifier (SRC), the Collaborative Representation Classifier (CRC), and the Two Phase Test Sample Sparse Representation (TPTSSR) classifier were proposed as classification tools that exploit sparse representation. Inspired by active learning techniques, this paper proposes an active Collaborative Representation Classifier that can be exploited by these supervised frameworks. The introduced Active Two Phase Collaborative Representation Classifier (ATPCRC) begins by estimating the label of the available unlabeled samples. At testing stage, based on the TPTSSR framework any test sample will have two representations that are calculated separately by using two different dictionaries. The first dictionary is composed of all samples having original labels. The second dictionary contains the whole dataset samples (original and predicted labels). The two kinds of class-wise reconstruction error are then fused in order to infer the label of the test sample. The proposal is validated on four public image datasets. The results shoe that the introduced ATPCRC can outperform the classic TPTSSR as well as several state-of-the-art approaches that use label and unlabeled data samples.