arXiv cs.AI· Pintu Kumar, Nandyala Hemachandra·· 12 小时前AI 评分18
面向图数据的标签噪声鲁棒算法 LENS：基于邻域标签分布
A Label Noise Robust Algorithm for Graph Data using Neighborhood Label Distribution
AI 导读
针对 GCN 依赖干净标签的问题，研究者提出端到端框架 LENS（Leave-one-out Entropy Node Screening），通过衡量移除某节点后邻域熵的下降来识别噪声节点，并结合重标注机制保留有效特征信息。该方法在节点分类上达到 SOTA 精度，运行速度比现有抗噪方法快一个数量级，且端到端计算开销极低。
正文 · 原文
Abstract:Graph Convolutional Networks (GCNs) have demonstrated exceptional performance on graph-structured data. However, their success heavily relies on the availability of clean labels, an assumption that often fails in real-world scenarios. The learnability of GCNs is intrinsically linked to the neighborhood structure of nodes, making it critical to understand how this structure changes in the presence of label noise. In this work, we investigate how the neighborhood label distribution of a node changes under label noise. We mathematically derive that label noise smooths this distribution, which inherently increases its entropy and severely decreases interclass distinguishability. To address this challenge, we propose Leave-one-out Entropy Node Screening (LENS), an end-to-end framework that identifies noisy nodes by measuring the reduction in neighborhood entropy when a specific node is removed. LENS further incorporates a relabeling mechanism to recover and preserve valuable feature information instead of entirely discarding the identified noisy nodes. Crucially, the end-to-end computational overhead of LENS is remarkably low. Extensive evaluations demonstrate that our algorithm achieves state-of-the-art node classification accuracy while executing an order of magnitude faster than existing noise-robust approaches.
| Subjects: | Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Statistics Theory (math.ST) |
| Cite as: | arXiv:2610.22998 [cs.LG] |
| (or arXiv:2610.22998v1 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2610.22998 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Pintu Kumar [view email]
[v1]
Fri, 16 Oct 2026 06:22:09 UTC (297 KB)
来源：arXiv cs.AI · arxiv.org