Tensor rank approximation with graph learning for robust multi-view clustering

Yongyi Yang

doi:10.1016/j.knosys.2026.115648

Tensor rank approximation with graph learning for robust multi-view clustering

2026-03-02

0

PRE

AI

DOI：10.1016/j.knosys.2026.115648

Original

Original request for help

Save

Abstract

En 中文

Graph learning is an important data processing technique in multi-view clustering, and currently graph learning is based on constructed similarity graphs, which are processed around separating the coherent information from the noise information. However, the research routes often suffer from insufficient mining of high-order correlations in consistent information and inadequate suppression of noise information. Therefore, we propose Tensor Rank Approximation with Graph Learning for robust multi-view clustering (TRAGL). Specifically, we construct a similarity graph on the Stiefel manifold, utilizing the mutual metrics between the sample points to maintain the intrinsic properties of the original data structure. Then, separate the consistency information and noise information of the similar graph. And construct tensors separately. We design a new tensor rank function Double Exponential Weighted Tensor Nuclear Norm (DETWTNN), to better constrain the consistency tensor and mine higher-order information. We design a new tensor sparse norm. We apply the dual constraints of tensor sparse norm and matrix norm to noise information. Finally, an optimization algorithm based on the alternating direction multiplier method (ADMM) is designed to solve the model efficiently. Extensive experimental results indicate that our proposed method outperforms most existing advanced multi-view clustering methods.

Keywords:

Graph Learning

Multi-View Clustering

Tensor Rank Approximation

Stiefel Manifold

Noise Suppression

Tensor rank approximation with graph learning for robust multi-view clustering

Abstract

Journal

Organization