AAAI-2025-Anchor Learning with Potential Cluster Constraints for Multi-view Clustering (ALPC)

#If you have any questions please contact (huibing.wang@dlmu.edu.cn) and (cyw@dlmu.edu.cn).

Introduction

In this paper, we propose a new Anchor Learning with Potential Cluster Constraints for Multi-view Clustering method (ALPC). ALPC integrates and optimizes anchor selection and anchor graph learning into a unified framework. Unlike previous approaches, ALPC imposes shared potential clustering semantic constraints along with orthogonal constraints on anchors. Aligning the clustering centers of data clusters with the clustering centers of anchors constrains anchors to be generated uniformly in data clusters, making anchors more representative as well as discriminative This allows for exploring inter-cluster diversity and intra-cluster consistency. ALPC truly realizes the relevance of anchors, and experiments demonstrate the effectiveness of the method.

<img width="555" height="450" alt="image" src="https://github.com/user-attachments/assets/19318fdc-bf13-450d-bc5f-aa3b5b49154b" />

Dependencies

• OS: Windows 10
• Matlab2020b
• Related measure can be obtained in .\tools.

Citation

If any part of our paper and repository is helpful to your work, please generously cite with:

@misc{chen2024anchorlearningpotentialcluster,
      title={Anchor Learning with Potential Cluster Constraints for Multi-view Clustering}, 
      author={Yawei Chen and Huibing Wang and Jinjia Peng and Yang Wang},
      year={2024},
      eprint={2412.16519},
      archivePrefix={arXiv},
      primaryClass={cs.CV},
      url={https://arxiv.org/abs/2412.16519}, 
}