Rényi Entropy Induced Efficient and Balanced One-Step Multi-View Clustering
Chuan Tang, Miaomiao Li, Jun Wang, Chang Tang, En Zhu, X Y Liu
National University of Defense Technology Changsha University Huazhong University of Science and Technology
阅读操作
确认中在文库中上传 PDF 后可生成中文音频讲解。
摘要与影响
Anchor-based multi-view clustering has attracted increasing interest due to its high clustering efficiency. Nevertheless, most existing methods construct a consensus anchor graph by considering only view-level contributions, neglecting the diverse significance of individual samples across different views. Additionally, these methods do not account for prior knowledge regarding the balance of sample distributions among clusters. To overcome these challenges, we propose a novel and efficient approach, named Rényi Entropy Induced Efficient and Balanced One-Step Multi-View Clustering (REIEBO). Our method introduces a sample-specific weighting strategy based on anchor self-representation learning, allowing the consensus anchor graph to better capture the sample importance disparities across views. To further enhance clustering performance, we introduce Rényi entropy as a regularization mechanism to dynamically adjust the clustering distribution by prior knowledge about sample balance. This adjustment is particularly beneficial for correctly assigning samples located near cluster boundaries. Moreover, an orthogonality constraint is imposed to directly yield final cluster assignments in a one-step fashion, effectively avoiding the information loss in conventional two-stage frameworks. Extensive experiments demonstrate that our method consistently outperforms state-of-the-art multi-view clustering methods. The code of REIEBO is publicly available athttps://github.com/tangchuan2000/REIEBO.
逐年被引趋势
关键指标
同类平均 = 1
同领域 · 同年份 · 同类型
Google Scholar 与 OpenAlex 的被引统计范围不同,数值存在差异属正常。
AI 辅助阅读
依据:摘要
可就本文提问;依据不足时会说明。
学术脉络
学科主题
计算机 / AIAdvanced Clustering Algorithms Research
Face and Expression Recognition · Advanced Graph Neural Networks