Multi-view Re-weighted Sparse Subspace Clustering with Intact Low-rank Space Learning

Multi-view Re-weighted Sparse Subspace Clustering with Intact Low-rank Space Learning

In this paper, we propose a new Multi-view Re-weighted Sparse Subspace Clustering with Intact Low-rank Space Learning (ILrS-MRSSC) method, trying to find a sparse representation of the complete space of information. Specifically, this method integrates the complementary information inherent in multiple angles of the data, learns a complete space of potential low-rank representation, and constructs a sparse information matrix to reconstruct the data. The correlation between multi-view learning and subspace clustering is strengthened to the greatest extent, so that the subspace representation is more intuitive and accurate. The optimal solution of the model is solved by the augmented lagrangian multiplier (ALM) method of alternating direction minimal. Experiments on multiple benchmark data sets verify the effec-tiveness of this method.