Adaptive weighted dictionary representation using anchor graph for subspace clustering

Abstract

Samples are commonly represented as sparse vectors in many dictionary representation algorithms. However, this method may result in loss of discriminatory information. Moreover, a redundant dictionary can increase the computational complexity of the algorithm. To tackle these challenges, we propose a novel method named Adaptive Weighted Dictionary Representation using Anchor Graph for Subspace Clustering (AWDR). First, AWDR constructs an anchor graph that encodes the classification information and establishes accurate connectivity components between anchors and clusters, thereby fully utilizing the discriminative information of the original samples. In addition, AWDR learns a complete-dictionary in the subspace to eliminate the noise and out-of-sample effects of the original sample space, while also improving computational efficiency. Finally, AWDR computes the coefficients for the samples in an adaptively weighted manner to find discriminative representation of the samples from the dictionary. Extensive experiments on real-world datasets demonstrate that our method is effective and efficient compared to the state-of-the-art methods.