Fast Tensor-Based Multi-View Clustering with Anchor Probability Transition Matrix
Keywords: Multi-view clustering, Fast clustering
Abstract: Multi-view clustering effectively integrates information from multiple data representations, yet current methods face key challenges. They often lack interpretability, obscuring how clusters are formed, and fail to fully leverage the complementary information across views, limiting clustering quality. Additionally, large-scale data introduces high computational demands, with traditional methods requiring extensive post-processing and manual tuning.To address these issues, we propose a novel multi-view clustering approach based on probability transition matrices. By selecting anchor points and constructing bipartite similarity graphs, we can capture the relationships between data points and anchors in different views and reduce computational complexity. Through probability matrices, we efficiently transfer cluster labels from anchors to samples, generating membership matrices without the need for post-processing. We further assemble these membership matrices into a tensor and apply a Schatten \(p\)-norm constraint to exploit complementary information across views, ensuring consistency and robustness. To prevent trivial solutions and ensure well-defined clusters, we incorporate nuclear norm-based regularization. Extensive experiments on various datasets confirm the effectiveness and efficiency of our method.
Supplementary Material: zip
Primary Area: unsupervised, self-supervised, semi-supervised, and supervised representation learning
Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.
Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2025/AuthorGuide.
Reciprocal Reviewing: I understand the reciprocal reviewing requirement as described on https://iclr.cc/Conferences/2025/CallForPapers. If none of the authors are registered as a reviewer, it may result in a desk rejection at the discretion of the program chairs. To request an exception, please complete this form at https://forms.gle/Huojr6VjkFxiQsUp6.
Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.
No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.
Submission Number: 1030
Loading