Graph Universal Adversarial Attacks: A Few Bad Actors Ruin Graph Learning Modelsdoi:10.24963/IJCAI.2021/458Xiao ZangYi XieJie ChenBo YuanInternational Joint Conferences on Artificial Intelligence Organization
与几个通用攻击基线相比,GUA达到了最高的ASR。在ASR和锚定集的大小之间存在一个折衷,作者发现非常小的尺寸就足以获得显著的成功攻击。 论文名称:Graph Universal Adversarial Attacks: A Few Bad Actors Ruin Graph Learning Models 论文链接:链接 AMiner已经对论文进行了AI精读,欢迎使用～ ...
Adversarial Attacks on Graph Classifiers via Bayesian OptimisationXingchen Wan (University of Oxford) · Henry Kenlay (University of Oxford) · Robin Ru (Oxford University) · Arno Blaas (University of Oxford) · Michael A Osborne (U Oxford) · Xiaowen Dong (University of Oxford) Robustness of ...
Exploratory Adversarial Attacks on Graph Neural Networks 依赖training loss的最大梯度的这种基于梯度的策略,在攻击GNN模型时候,可能不会产生一个好的结果。 原因在于图结构的离散的特点。 ⇓ \Downarrow ⇓ 我们可不可以推导出一种有效的方法,来选择攻击GNN的扰动? 我们提出一种新颖的explorat... ...
Watermarking Graph Neural Networks based on Backdoor Attacks, 📝arXiv Robustness of Graph Neural Networks at Scale, 📝NeurIPS, Code Generalization of Neural Combinatorial Solvers Through the Lens of Adversarial Robustness, 📝NeurIPS Graph Universal Adversarial Attacks: A Few Bad Actors Ruin Graph ...
2023 Preventing Attacks in Interbank Credit Rating with Selective-aware Graph Neural Network IJCAI 2023 Link Link 2023 Fighting against Organized Fraudsters Using Risk Diffusion-based Parallel Graph Neural Network IJCAI 2023 Link Link 2023 Diga: Guided Diffusion Model for Graph Recovery in Anti-Money La...
2204). In addition, graph data may also be subject to malicious attacks, such as adversarial attacks, which are designed to mislead the performance of a graph pooling method by adding, deleting, or modifying nodes and edges in the graph. Therefore, robust graph pooling methods need to have ...
Adversarial Attacks on Graph Neural Networks via Meta Learning Daniel Zügner, Stephan Günnemann ICLR 2019 4.10 元学习 Learning Steady-States of Iterative Algorithms over Graphs Hanjun Dai, Zornitsa Kozareva, Bo Dai, Alex Smola, Le Song
Marinka Zitnik Abstract Networks—or graphs—are universal descriptors of systems of interacting elements. In biomedicine and healthcare, they can represent, for example, molecular interactions, signalling pathways, disease co-morbidities or healthcare systems. In this Perspective, we posit that represent...
Adversarial Attacks on Graph Neural Networks via Meta Learning Daniel Zügner, Stephan Günnemann ICLR 2019 4.10 元学习 Learning Steady-States of Iterative Algorithms over Graphs Hanjun Dai, Zornitsa Kozareva, Bo Dai, Alex Smola, Le Song