.png){kind=icon}
返回
Multi-view subspace clustering via tensor nuclear norm factorization
.png){kind=icon}
PRE
AI
DOI：10.1016/j.patcog.2025.112969.png)
.png){kind=icon}
原文求助.png){kind=icon}
分享.png){kind=icon}
收藏
摘要
En 中文
Multi-view clustering is a powerful technique that leverages complementary information from multiple distinct views to partition samples into clusters. Among various methods, low-rank tensor-based multi-view subspace clustering (MVSC) methods have gained prominence due to their ability to effectively capture high-order correlations in multi-view data, thereby enhancing the extraction of complementary information across views. However, these methods often rely on computing singular value decomposition (SVD) of large-scale matrices to enforce low-rank constraints, which incurs significant computational costs. To alleviate this issue, we propose a tensor factorization strategy that employs the Frobenius norm regularization to impose low-rankness. Based on this, we introduce a novel MVSC model via tensor nuclear norm factorization (TNNF-MVSC), which not only reduces time consumption but also maintains impressive clustering accuracy. Furthermore, we propose a novel incomplete MVSC model (TNNF-IMVSC) by extending the tensor factorization strategy to handle incomplete multi-view data. Extensive experiments demonstrate that TNNF-MVSC and TNNF-IMVSC outperform state-ofthe-art methods in both efficiency and effectiveness. The demo code for our methods is publicly available at https://github.com/TH-Yan/TNNF-MVSC.
Keyword:
Multi-view subspace clustering
Low-rank representation
Singular value decomposition
Tensor factorization
Tensor nuclear norm
Incomplete view
期刊
IF:
7.6
论文数:
1.3W
被引数:
4.5W
机构
引用论文
Joint local smoothness and low-rank tensor representation for robust multi-view clustering.png)
PATTERN RECOGNITION
PATTERN RECOGNITIONIF7.6
PREAI
分享收藏Multi-view subspace clustering via simultaneously learning the representation tensor and affinity matrixPATTERN RECOGNITION
PATTERN RECOGNITIONIF7.6
PREAI
分享收藏分享收藏Tensor robust principal component analysis via dual l quasi-norm sparse constraints张量鲁棒主成分分析：基于双l准范数的稀疏约束
OAAI
OAAI分享收藏Tensorized latent representation with automatic dimensionality selection for multi-view clustering张量化的潜在表示与自动维度选择用于多视图聚类PATTERN RECOGNITION
PATTERN RECOGNITIONIF7.6
PREAI
分享收藏
查看更多内容