mlsanityData to 14 Oct 2026

IDEA: Invariant Defense for Graph Adversarial Robustness

arXiv 2305.15792 ↗First posted 25 May 2023cs.LG cs.CRcited by 4

Not on the map. Why?

Abstract (arXiv)

Despite the success of graph neural networks (GNNs), their vulnerability to adversarial attacks poses tremendous challenges for practical applications. Existing defense methods suffer from severe performance decline under unseen attacks, due to either limited observed adversarial examples or pre-defined heuristics. To address these limitations, we analyze the causalities in graph adversarial attacks and conclude that causal features are key to achieve graph adversarial robustness, owing to their determinedness for labels and invariance across attacks. To learn these causal features, we innovatively propose an Invariant causal DEfense method against adversarial Attacks (IDEA). We derive node-based and structure-based invariance objectives from an information-theoretic perspective. IDEA ensures strong predictability for labels and invariant predictability across attacks, which is provably a causally invariant defense across various attacks. Extensive experiments demonstrate that IDEA attains state-of-the-art defense performance under all five attacks on all five datasets. The implementation of IDEA is available at https://anonymous.4open.science/r/IDEA.

Key facts. Cited by 4 papers in the map (data through 14 Oct 2026). Builds on 52 of its 76 references that are in the map. First posted on arXiv 25 May 2023.

Builds on

52 of 76 references.

References · 52

  1. 2016Semi-Supervised Classification with Graph Convolutional Networkscited by 11,361
  2. 2017Graph Attention Networkscited by 8,923
  3. 2017Towards Deep Learning Models Resistant to Adversarial Attackscited by 6,415
  4. 2013Auto-Encoding Variational Bayescited by 4,053
  5. 2019Invariant Risk Minimizationcited by 1,854
  6. 2020LightGCN: Simplifying and Powering Graph Convolution Network for Recommendationcited by 1,139
  7. 2016Deep Variational Information Bottleneckcited by 930
  8. 2020Out-of-Distribution Generalization via Risk Extrapolation (REx)cited by 742
  9. 2019GraphSAINT: Graph Sampling Based Inductive Learning Methodcited by 618
  10. 2021Towards Out-Of-Distribution Generalization: A Surveycited by 436
  11. 2018Adversarial Attacks on Neural Networks for Graph Datacited by 397
  12. 2020Graph Structure Learning for Robust Graph Neural Networkscited by 371
38 more
  1. 2017Defense against Adversarial Attacks Using High-Level Representation Guided Denoisercited by 329
  2. 2020Environment Inference for Invariant Learningcited by 306
  3. 2020The Risks of Invariant Risk Minimizationcited by 262
  4. 2019Adversarial Attacks on Graph Neural Networks via Meta Learningcited by 261
  5. 2020CLUB: A Contrastive Log-ratio Upper Bound of Mutual Informationcited by 235
  6. 2020GNNGuard: Defending Graph Neural Networks against Adversarial Attackscited by 195
  7. 2020A Self-supervised Approach for Adversarial Robustnesscited by 180
  8. 2018Adversarial Attack and Defense on Graph Data: A Surveycited by 167
  9. 2022Handling Distribution Shifts on Graphs: An Invariance Perspectivecited by 164
  10. 2022Learning Causally Invariant Representations for Out-of-Distribution Generalization on Graphscited by 144
  11. 2020Node Similarity Preserving Graph Convolutional Networkscited by 112
  12. 2021Robustness of Graph Neural Networks at Scalecited by 100
  13. 2019Graph Adversarial Training: Dynamically Regularizing Based on Graph Structurecited by 93
  14. 2021Invariant Information Bottleneck for Domain Generalizationcited by 86
  15. 2020DeepRobust: A PyTorch Library for Adversarial Attacks and Defensescited by 78
  16. 2020Learning Invariant Representations and Risks for Semi-supervised Domain Adaptationcited by 69
  17. 2021TDGIA:Effective Injection Attacks on Graph Neural Networkscited by 67
  18. 2021Elastic Graph Neural Networkscited by 67
  19. 2019Certifiable Robustness to Graph Perturbationscited by 62
  20. 2022Understanding and Improving Graph Injection Attack by Promoting Unnoticeabilitycited by 62
  21. 2019Certifiable Robustness and Robust Training for Graph Convolutional Networkscited by 59
  22. 2020Robust Optimization as Data Augmentation for Large-scale Graphscited by 58
  23. 2022ZIN: When and How to Learn Invariance Without Environment Partition?cited by 58
  24. 2020Efficient Robustness Certificates for Discrete Data: Sparsity-Aware Randomized Smoothing for Graphs, Images and Morecited by 56
  25. 2020Scalable Attack on Graph Data by Injecting Vicious Nodescited by 50
  26. 2021Single Node Injection Attack against Graph Neural Networkscited by 45
  27. 2020Adversarial Attacks and Defenses on Graphs: A Review, A Tool and Empirical Studiescited by 42
  28. 2020A Survey of Adversarial Learning on Graphscited by 42
  29. 2023Are Defenses for Graph Neural Networks Robust?cited by 38
  30. 2022Reliable Representations Make A Stronger Defender: Unsupervised Structure Refinement for Robust GNNcited by 34
  31. 2019Adversarial Training Methods for Network Embeddingcited by 26
  32. 2023Revisiting Robustness in Graph Machine Learningcited by 19
  33. 2021Removing Adversarial Noise in Class Activation Feature Spacecited by 18
  34. 2022When Does Group Invariant Learning Survive Spurious Correlations?cited by 18
  35. 2023Collective Robustness Certificates: Exploiting Interdependence in Graph Neural Networkscited by 16
  36. 2022Rumor Detection with Self-supervised Learning on Texts and Social Graphcited by 15
  37. 2020Adversarial Immunization for Certifiable Robustness on Graphscited by 14
  38. 2022Spectral Adversarial Training for Robust Graph Neural Networkcited by 12

50 most cited of 52 All 52 (JSON)

Built on by

Cited by year

Citing papers per year: 2023: 3; 2024: 1.

4 papers.

  1. 2023Does Invariant Graph Learning via Environment Augmentation Learn Invariance?cited by 42
  2. 2023Exploring Causal Learning through Graph Neural Networks: An In-depth Reviewcited by 8
  3. 2024A Survey of Out-of-distribution Generalization for Graph Machine Learning from a Causal Viewcited by 3
  4. 2023DAD++: Improved Data-free Test Time Adversarial Defensecited by 0