.png){kind=icon}
Return
Tensorized topological manifold for multiple kernel clustering
.png){kind=icon}
PRE
AI
DOI：10.1016/j.ins.2026.123188.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Multiple kernel clustering (MKC) has gained significant attention due to its capacity to efficiently handle nonlinear information. However, existing MKC methods often overlook heterogeneous neighbor relations and fail to fully capture high-order complementary information across kernels. Consequently, these limitations impede its widespread adoption in practical applications. To address these limitations, we propose a novel clustering method termed Tensorized Topological Manifold for Multiple Kernel Clustering (i.e., TTMMKC). Specifically, TTMMKC constructs a kernel topological graph by adopting a manifold topological structure as a means to measure similarities between instances instead of the Euclidean structure to explore the local neighbor propagation structure in Hilbert space. Moreover, TTMMKC integrates multiple kernel topological graphs into a third-order tensor and utilizes the tensor nuclear norm to explore complementary consistency information between heterogeneous kernel topological graphs. Experimental results on benchmark datasets across various scenarios demonstrate that TTMMKC effectively leverages higher-order topological structures to accurately capture the intrinsic propagation information of data in Hilbert space.
Keywords:
Multiple kernel clustering
Graph learning
Tensor learning
Topological manifold
Journal
IF:
6.8
Papers:
553
Citations:
6.2W
Organization
View more organizations
Cited Papers
Multi-view clustering via efficient representation learning with anchors.png)
PATTERN RECOGNITION
PATTERN RECOGNITIONIF7.6
PREAI
ShareSaveShareSaveShareSave
ShareSave View more