Subspace clustering aims to group data points into multiple clusters of whicheach corresponds to one subspace. Most existing subspace clustering methodsassume that the data could be linearly represented with each other in the inputspace. In practice, however, this assumption is hard to be satisfied. Toachieve nonlinear subspace clustering, we propose a novel method which consistsof the following three steps: 1) projecting the data into a hidden space inwhich the data can be linearly reconstructed from each other; 2) calculatingthe globally linear reconstruction coefficients in the kernel space; 3)truncating the trivial coefficients to achieve robustness andblock-diagonality, and then achieving clustering by solving a graph Laplacianproblem. Our method has the advantages of a closed-form solution and capacityof clustering data points that lie in nonlinear subspaces. The first advantagemakes our method efficient in handling large-scale data sets, and the secondone enables the proposed method to address the nonlinear subspace clusteringchallenge. Extensive experiments on five real-world datasets demonstrate theeffectiveness and the efficiency of the proposed method in comparison with tenstate-of-the-art approaches regarding four evaluation metrics.
展开▼
