Proceedings

Resisting Label Drift: Real-Time Multi-View Clustering with Semantic Consistency

Qi Liu, Suyuan Liu, Hao Tan, Yangfan Du, Bowen Zhang, Wenpeng Lu, Xinwang Liu

Download BibTeX Cite

Proceedings of the Thirty-Fifth International Joint Conference on Artificial Intelligence. Main Track. Pages 2870-2878.

https://doi.org/10.24963/ijcai.2026/319

Keywords

  • Machine Learning: Clustering
  • Data Mining: Mining data streams
  • Machine Learning: Multi-view learning
  • Machine Learning: Online learning

Abstract

Real-time clustering of dynamic multi-view data streams is a critical yet challenging task in open-world applications. While several methods have been proposed to address this task, most of them extract features incrementally but fail to output instant clustering results for the current batch. In addition, they neglect semantic consistency, causing the cluster labels of identical concepts to drift unpredictably due to independent processing. To address these limitations, we propose a real-time semantic consistent incremental multi-view clustering framework. Specifically, we constructs a compact historical knowledge base via an adaptive diversity-aware selection mechanism, which guides the clustering of incoming data, enabling immediate inference without accessing the full history. Furthermore, we introduce a semantic alignment strategy based on consensus centers to ensure robust label consistency over time. Extensive experiments demonstrate the effectiveness and efficiency of our proposed method.

Qi Liu, Suyuan Liu, Hao Tan, Yangfan Du, Bowen Zhang, Wenpeng Lu, and Xinwang Liu. 2026. Resisting Label Drift: Real-Time Multi-View Clustering with Semantic Consistency. In Proceedings of the Thirty-Fifth International Joint Conference on Artificial Intelligence, IJCAI-26. IJCAI, 2870–2878. https://doi.org/10.24963/ijcai.2026/319
Welcome Future Conferences AI Journal Division IJCAI Proceedings About IJCAI IJCAI Awards