An Efficient and Robust Spectral Clustering Approach for Subspace Classification

Traditional low-rank representation (LRR)and variant algorithms based on LRR have been proved as effective methods for subspace clustering tasks in machine learning. However, the majority of the existing LRR-related algorithms relax the rank minimization problem by the trace norm minimization. It is rarely accurate to get the optimal solution guaranteed by the definition. Meanwhile, these methods can not well reveal the internal structure of the classification. In this paper, we propose an efficient and robust subspace clustering method to overcome these drawbacks. A robust self-representation coefficient matrix is learned by utilizing the Schatten-
$p$
norm instead of the conventional rank function. Besides, the strongest block-diagonal structure of the coefficient representation matrix is enhanced by learning and optimizing the co-association matrix with the soft label of clustering results simultaneously in an unified framework. The affinity graphs constructed in this paper can clearly reveal the intrinsic structures of the data sets. Extensive experiments on the real data sets demonstrate that our novel proposed method can perform more effective than the state-of-the-art methods.