Robust Multi-view Spectral Clustering with Smoothed Anchor Graph Learning

Multi-view clustering has garnered significant attention due to its ability to leverage information from multiple perspectives. Enhancing clustering performance has long been a key research focus in this field. However, due to noise interference in real-world data, existing methods often struggle to balance clustering accuracy and robustness. Additionally, the high time complexity associated with constructing full-sample similarity graphs severely limits the applicability of these algorithms in large-scale data scenarios. To address this issue, we propose a Robust Multi-view Spectral Clustering method based on Smoothed Anchor Graph Learning (RMSCL). This method integrates consensus graph learning and spectral embedding into a unified framework, effectively obtaining clustering results without introducing additional computational steps. By replacing the full-sample graph with an anchor graph, it significantly reduces computational complexity. Furthermore, it incorporates graph filtering to mitigate the impact of noise on clustering performance. Extensive experiments demonstrate that RMSCL outperforms state-of-the-art baseline methods in both clustering performance and robustness.