点击次数：

论文类型：会议论文

发表时间：2017-01-01

收录刊物：Scopus、EI、CPCI-S

卷号：10559

页面范围：461-470

关键字：Subspace clustering; Multiplicative noise

摘要：Traditional subspace clustering models generally adopt the hypothesis of additive noise, which, however, dose not always hold. When it comes to multiplicative noise corruption, these models usually have poor performance. Therefore, we propose a novel model for robust subspace clustering with multiplicative noise corruption to alleviate this problem, which is the key contribution of this work. The proposed model is evaluated on the Extend Yale B and MNIST datasets and the experimental results show that our method achieves favorable performance against the state-of-the-art methods.