Sep 2026· IEEE Transactions on Image Processing· Vol 35, pp. 10682-10696· 0 citations· 53 references
MedicineComputer Science
TL;DR
This paper proposes a novel framework termed Few-shot Anchor-guided Multi-view Clustering with Pearson Correlation (FAMC-PC), which establishes a tuning-free design, substantially reducing the computational burden and labor costs associated with manual hyper-parameter tuning.
Abstract
Multi-view clustering (MVC) has emerged as a powerful paradigm for integrating heterogeneous data representations. However, existing multi-view clustering methods typically encounter four specific bottlenecks: the cubic computational explosion inherent to spectral graph methods, the degradation of Euclidean distance in high-dimensional spaces, extreme label scarcity in practical semi-supervised applications, and a heavy reliance on costly manual hyper-parameter tuning. To address these challenges, this paper proposes a novel framework termed Few-shot Anchor-guided Multi-view Clustering with Pearson Correlation (FAMC-PC). Unlike traditional approaches, FAMC-PC introduces a statistical Pearson Correlation metric to construct bipartite anchor graphs, capturing intrinsic structural directionality more effectively than relying on purely distance-based measures. We further propose a unified Non-negative Matrix Factorization (NMF) model that seamlessly integrates consensus graph fusion with sparse few-shot constraints. This mechanism anchors latent representations to scarce labeled data, bridging the gap between unsupervised structure learning and supervised classification without requiring extensive annotations. Notably, FAMC-PC establishes a tuning-free design, substantially reducing the computational burden and labor costs associated with manual hyper-parameter tuning. Extensive experiments on six benchmark datasets demonstrate that FAMC-PC achieves competitive clustering performance and high efficiency compared to nine state-of-the-art baselines, offering a significant advantage in terms of label efficiency. The source code is available at https://github.com/LstinWh/FAMC-PC
The Dual-tOpology learning with adapTive Anchors (DOTA) is proposed, which not only learns the sample-anchor relationship but also preserves the topology structure among anchors, significantly enhancing the discriminability of learned representation while preserving the underlying data manifold.
Cheng-Long Zhang, Chao Zhang, Jun-Hao Zhang et al.· Proceedings of the Thirty-Fi...· 0 citations
This work proposes a novel deep constrained multi-view clustering framework, namely SeSCE, which encodes view embeddings into a spherical space, leveraging pairwise constraints to maximize intra-class compactness and inter-class separability and introduces a confidence-aware pseudo-constraint mining mechanism.
Jun Wang, Zhenglai Li, Chuan Tang et al.· IEEE Transactions on Pattern...· 0 citations
Multi-view Clustering via Manifold Decomposition is proposed, which directly infers cluster labels from multi-view data without explicit similarity graph construction or anchor selection, and reformulates multi-view clustering as a unified multi-view regression problem, where cluster labels are optimized as model varia...
Xiao-Wei Zhao, Xin-Yue Kou, Yan Chen et al.· Proceedings of the Thirty-Fi...· 0 citations
An innovative MVC method utilizing orthogonal non-negative anchor tensor factorization, termed ATFMC, which employs a shared-nearest-neighbor density peaks clustering algorithm, which integrates cross-view features to select high-quality anchors and achieves superior clustering performance.
Jia-Yi Wang, Ming Yang, Jing-Yu Wang et al.· ACM Transactions on Intellig...· 0 citations
A Correntropy-guided Matrix Factorization and Tensor Graph Learning (CMTGL) framework for robust semi-supervised multi-view clustering that incorporates partial label information into a shared low-dimensional representation through constrained low-rank matrix factorization and employs the maximum correntropy criterion...
Lin Hu, Song Jiang, Xiu Liu et al.· Symmetry· 0 citations
This paper proposes salient-residual decoupled multi-view learning for clustering, SRDMVC, introducing a novel decomposition-fusion iterative optimization, which separates the feature space into a salient space and a residual subspace effectively and fuses them using a novel attention mechanism.
Gao-Kai Wang, Yazhou Ren, Feng-Yu Zhang et al.· Proceedings of the Thirty-Fi...· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.