Research map / Graph learning / Graph self-supervised learning
GADBench: Revisiting and Benchmarking Supervised Graph Anomaly Detection
Jianheng Tang, Fengrui Hua, Ziqi Gao, Peilin Zhao, Jia Li
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 tree ensembles with simple neighborhood aggregation instead of GNN methods | Supervised graph anomaly detection · 10 real-world GAD datasets, fully-supervised | AUPRC: +12.9% AUPRC vs BWGNN | Improved |
| Use tree ensembles with simple neighborhood aggregation instead of GNN methods | Supervised graph anomaly detection · 10 real-world GAD datasets, semi-supervised | AUPRC: +8.0% AUPRC vs GHRN | Improved |
| Tune model hyperparameters with random search | Standard GNNs · GADBench datasets | AUPRC: GraphSAGE average AUPRC improvement of 10.4% | Improved |
| Tune specialized GNN hyperparameters | Specialized GNNs · GADBench datasets | AUPRC | Improved |
| Add Neighborhood Averaging (ablation) | XGBoost · 10 datasets, semi-supervised | AUPRC: 37.5% -> 38.9% | Improved |
| Add Neighborhood Averaging (ablation) | XGB-Graph · 10 datasets, semi-supervised | AUPRC: 42.91 -> 44.30 | Improved |
| Add Neighborhood Averaging (ablation) | XGB-Graph · 10 datasets, fully-supervised | AUPRC: 61.09 -> 59.58 | Hurt |
| Add Neighborhood Averaging (ablation) | RF-Graph · 10 datasets, semi-supervised | AUPRC: 43.37 -> 45.14 | Improved |
| Add Neighborhood Averaging (ablation) | RF-Graph · 10 datasets, fully-supervised | AUPRC: 59.65 -> 58.54 | Hurt |
| Add Neighborhood Averaging (ablation) | GIN · 10 datasets, semi-supervised | AUPRC: 30.26 -> 33.91 | Improved |
| Increase neighbor aggregation depth from 0 to 2 layers (ablation) | XGB-Graph and RF-Graph · Most datasets | performance | Improved |
| Increase neighbor aggregation depth beyond 2 layers (ablation) | XGB-Graph and RF-Graph · Most datasets | performance | No clear effect |
Closest papers by meaning
- ARC: A Generalist Graph Anomaly Detector with In-Context Learning (NeurIPS 2024 · similarity 0.91)
- Generative Semi-supervised Graph Anomaly Detection (NeurIPS 2024 · similarity 0.91)
- Open-Set Graph Anomaly Detection via Normal Structure Regularisation (ICLR 2025 · similarity 0.90)
- Graph Anomaly Detection as Dynamical Transport: Training-Free Scoring via Empirical Bayes (NeurIPS 2026 · similarity 0.90)
- ADA-GAD: Anomaly-Denoised Autoencoders for Graph Anomaly Detection (AAAI 2024 · similarity 0.90)
- Unifying Unsupervised Graph-Level Anomaly Detection and Out-of-Distribution Detection: A Benchmark (ICLR 2025 · similarity 0.89)
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.
