Research map / Adversarial robustness and security / Attacks on text and graphs
Are Defenses for Graph Neural Networks Robust?
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| replace non-adaptive transfer attacks with adaptive attacks | 7 GNN defenses (Jaccard-GCN, SVD-GCN, ProGNN, GRAND, GNNGuard, RGCN, Soft-Median-GDC) · Cora-ML and Citeseer | (R)AUC robustness: -40% on average | Hurt |
| use defenses instead of undefended tuned GCN | GNN defenses · Cora-ML and Citeseer, global and local | adversarial accuracy under adaptive attack | No clear effect |
| use GRAND or Soft-Median-GDC instead of GCN | GCN · Cora-ML and Citeseer | adversarial accuracy | Mixed |
| use SVD-GCN defense instead of GCN | GCN · Cora-ML | adversarial accuracy: 24%/9% | Hurt |
| higher clean accuracy of defenses | GNN defenses · Cora-ML global attacks | RAUC | Helped |
Closest papers by meaning
- If You Want to Be Robust, Be Wary of Initialization (NeurIPS 2024 · similarity 0.92)
- Bounding the Expected Robustness of Graph Neural Networks Subject to Node Feature Attacks (ICLR 2024 · similarity 0.92)
- A Simple and Yet Fairly Effective Defense for Graph Neural Networks (AAAI 2024 · similarity 0.92)
- Value at Adversarial Risk: A Graph Defense Strategy against Cost-Aware Attacks (AAAI 2024 · similarity 0.91)
- Adversarial Robust Generalization of Graph Neural Networks (ICML 2025 · similarity 0.91)
- Robust Graph Neural Networks via Unbiased Aggregation (NeurIPS 2024 · similarity 0.90)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.