arXiv cs.AI· Zhichao Hou, Xitong Zhang, Wei Wang, Charu C. Aggarwal, Xiaorui Liu·· 12 小时前AI 评分20
利用有向图中的信任信号：重新定义图学习的鲁棒性
Harnessing Trust in Directed Graphs: Redefining Robustness of Graph Learning
AI 导读
研究重新定义了有向图上 GNN 的鲁棒性，指出既有鲁棒图神经网络研究主要针对无向图，忽略了链接方向中蕴含的信任信号。作者从攻击侧提出更贴近真实的有向图攻击设定，从防御侧提出一个可插拔的消息传递框架，避免虚假的鲁棒感。该框架与现有防御策略结合后，在迁移攻击和自适应攻击下取得 SOTA 鲁棒性能，同时保持较强的干净准确率。
正文 · 原文
Abstract:Existing research on robust Graph Neural Networks (GNNs) focuses predominantly on undirected graphs, overlooking the trustworthiness inherent in the link directions of directed graphs. In this work, we analyze the limitations of current approaches from both the attack and defense perspectives and redefine the robustness of GNNs in directed graphs. From the attack side, we introduce a more realistic directed graph attack setting that addresses the shortcomings of existing attacks. From the defense side, we propose a simple yet effective message-passing framework that serves as a plug-in layer to strengthen GNN robustness while avoiding a false sense of robustness. Our findings show that the trust signals encoded in directed graphs can be harnessed to substantially improve the robustness of GNNs. When combined with existing defense strategies, the framework achieves state-of-the-art robust performance alongside strong clean accuracy against both transfer and adaptive attacks. These results point to a novel and promising direction for strengthening the robustness of graph deep learning.
| Subjects: | Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Cryptography and Security (cs.CR) |
| Cite as: | arXiv:2306.02002 [cs.LG] |
| (or arXiv:2306.02002v2 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2306.02002 arXiv-issued DOI via DataCite |
Submission history
From: Zhichao Hou [view email]
[v1]
Sat, 3 Jun 2023 04:56:04 UTC (507 KB)
[v2]
Fri, 16 Oct 2026 05:26:28 UTC (1,603 KB)
来源：arXiv cs.AI · arxiv.org