Research map / Adversarial robustness and security / Attacks on text and graphs
UAG: Uncertainty-aware Attention Graph Neural Network for Defending 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 |
|---|---|---|---|
| Use UAG with both model and data uncertainty | GCN-based UAG · Cora, Citeseer, and Pubmed under Random, DICE, and Mettack attacks | accuracy | Helped |
| Combine model and data uncertainty instead of using either alone (ablation) | UAG · Cora, Citeseer, and Pubmed under diverse attack ratios | accuracy | Helped |
| Use data uncertainty based on node diversity to guide attention (ablation) | UAG · Cora, Citeseer, and Pubmed under DICE attack | accuracy | Helped |
| Train data uncertainty with both labeled-node loss L1 and unlabeled-node loss L2 (ablation) | UAG data uncertainty branch · Loss Design Benefits experiment | accuracy | Helped |
| Increase model or data uncertainty (ablation) | UAG · Accuracy and Uncertainty experiment | accuracy | Hurt |
| Increase the attention impact of uncertain nodes using larger ατ or smaller βτ (ablation) | UAG · Higher attack ratios | accuracy | Helped |
| Set ατ too large or βτ too small (ablation) | UAG · Parameter Analysis | accuracy | Hurt |
| Use larger static edge weights (ablation) | GNN node aggregation under attack · Static Edge Weights experiment | accuracy | Hurt |
Closest papers by meaning
- Value at Adversarial Risk: A Graph Defense Strategy against Cost-Aware Attacks (AAAI 2024 · similarity 0.87)
- If You Want to Be Robust, Be Wary of Initialization (NeurIPS 2024 · similarity 0.87)
- Adversarial Robust Generalization of Graph Neural Networks (ICML 2025 · similarity 0.86)
- A Simple and Yet Fairly Effective Defense for Graph Neural Networks (AAAI 2024 · similarity 0.86)
- Bounding the Expected Robustness of Graph Neural Networks Subject to Node Feature Attacks (ICLR 2024 · similarity 0.86)
- Accurate and Scalable Estimation of Epistemic Uncertainty for Graph Neural Networks (ICLR 2024 · similarity 0.86)
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.