Research map / Graph learning / GNN architectures
GNNGuard: Defending Graph Neural Networks against Adversarial Attacks
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 |
|---|---|---|---|
| Add GNNGuard to GCN | GCN · Cora, Nettack-Di | accuracy: 0.525 -> 0.705 | Helped |
| Add GNNGuard to GCN | GCN · Citeseer, Nettack-Di | accuracy: 0.615 -> 0.720 | Helped |
| Add GNNGuard to GCN | GCN · ogbn-arxiv, Nettack-Di | accuracy: 0.370 -> 0.425 | Helped |
| Add GNNGuard to GCN | GCN · DP, Nettack-Di | accuracy: 0.395 -> 0.430 | Helped |
| Add GNNGuard to GIN | GIN · Cora, Nettack-In | accuracy: 0.635 -> 0.775 | Helped |
| Add GNNGuard to GIN | GIN · Citeseer, Nettack-In | accuracy: 0.675 -> 0.845 | Helped |
| Add GNNGuard to GIN | GIN · ogbn-arxiv, Nettack-In | accuracy: 0.620 -> 0.710 | Helped |
| Add GNNGuard to GIN | GIN · DP, Nettack-In | accuracy: 0.605 -> 0.695 | Helped |
| Add edge pruning (ablation) | GNNGuard · ogbn-arxiv, Nettack-Di | accuracy: 0.350 -> 0.425 (GCN); 0.315 -> 0.520 (GAT); 0.540 -> 0.640 (GIN); 0.565 -> 0.635 (JK-Net); 0.305 -> 0.375 (GraphSAINT) | Helped |
| Add layer-wise graph memory (ablation) | GNNGuard · ogbn-arxiv, Nettack-Di | accuracy: 0.405 -> 0.425 (GCN); 0.475 -> 0.520 (GAT); 0.610 -> 0.640 (GIN); 0.625 -> 0.635 (JK-Net); 0.360 -> 0.375 (GraphSAINT) | Helped |
| Add GNNGuard to GIN | GIN · Cora, Mettack | accuracy: 0.702 -> 0.722 | Helped |
| Add GNNGuard to GIN | GIN · Citeseer, Mettack | accuracy: 0.638 -> 0.711 | Helped |
Closest papers by meaning
- Value at Adversarial Risk: A Graph Defense Strategy against Cost-Aware Attacks (AAAI 2024 · similarity 0.91)
- A Simple and Yet Fairly Effective Defense for Graph Neural Networks (AAAI 2024 · similarity 0.90)
- CEGA: A Cost-Effective Approach for Graph-Based Model Extraction and Acquisition (ICML 2025 · similarity 0.90)
- Grimm: A Plug-and-Play Perturbation Rectifier for Graph Neural Networks Defending Against Poisoning Attacks (AAAI 2025 · similarity 0.89)
- Adversarial Robust Generalization of Graph Neural Networks (ICML 2025 · similarity 0.89)
- When Witnesses Defend: A Witness Graph Topological Layer for Adversarial Graph Learning (AAAI 2025 · similarity 0.88)
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.