这篇文章研究的内容是:通过改变图的拓扑结构,来影响分类器模型的预测结果,进而研究分类器模型到底学到了什么,并且有效提升模型的鲁棒性。 这篇文章首先提出了一种基于强化学习的攻击方法,该方法只需要分类器模型的预测标签,就可以学习到有效的攻击策略;另外在给定模型梯度信息的情况下,提出了一种叫“梯度算法”的攻击方法。在给定模型的预测置信
Inspired by the practical importance of graph structured data, link prediction, one of the most frequently applied tasks on graph data, has garnered considerable attention in recent years, and they have been widely applied in item recommendation, privacy inference attack, knowledge graph completion, ...
ICML 2018 的一篇论文《Adversarial Attack on Graph Structured Data》试图探讨对 GNN 网络进行对抗攻击,并尝试了多种算法。 虽然深度学习在计算机视觉、语音识别、NLP等领域中的很多任务都取得了显著的突破性成果,深度神经网络模型也越来越完善,但是这些技术是否真正成熟,产品是否足够安全、可靠?这些将成为以后越来越被...
An Efficient Adversarial Attack on Graph Structured Data 📝IJCAI Workshop Model Algorithm Surrogate Target Task Target Model Baseline Metric Dataset Practical Adversarial Attacks on Graph Neural Networks 📝ICML Workshop Model GC-RWCS Algorithm Greedy Surrogate Target Task Node Classification Tar...
Since the GNN is trained based on the node's features and the connection relationship between the nodes, an attacker can add a small amount of error information to the training data in order to attack the GNN. Typically, adversarial attacks can be categorized into three types: white-box attac...
Dai, H., et al.: Adversarial attack on graph structured data. In: CoRR abs/1806.02371 (2018).arXiv: 1806.02371 Dalvi, N., et al.: Adversarial classification. In: Proceedings of the Tenth ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, KDD 2004, pp. 99–108. ...
Although similar strategies have been used in classifiers, graph-structured data45,46, and physical models47, no work has yet connected these strategies to sample multidimensional potential energy landscapes. In this framework, an adversarial attack maximizes the uncertainty in the property under ...
safe-graph/graph-adversarial-learning-literature Star842 Code Issues Pull requests A curated list of adversarial attacks and defenses papers on graph-structured data. securitymachine-learningdata-miningdeep-learninggraph-algorithmssurveyawesome-listgraph-datagraph-attackliterature-reviewadversarial-machine-learning...
[11] Adversarial attack on graph neural networks as an influence maximization problem. WSDM'22[12] Adversarial attack on graph structured data. PMLR'18[13] Fast gradient attack on network embedding. arXiv'18[14] Adversarial examples on graph data: Deep insights into attack and defense. ar...
On the Adversarial Robustness of Mixture of Experts计算成本不变但扩大模型规模,从而提高鲁棒性 Towards Consistency in Adversarial Classification Adversarially Robust Learning: A Generic Minimax Optimal Learner... Adv-Attribute: Inconspicuous and Transferable Adversarial Attack on... ...