Research map / Graph learning / Graph self-supervised learning
OWLEYE: ZERO-SHOT LEARNER FOR CROSSDOMAIN GRAPH DATA ANOMALY DETECTION
Lecheng Zheng, Dongqi Fu, Zihao Li, Jingrui He
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 OwlEye instead of ARC (zero-shot) | graph anomaly detection · 8 test graphs average | AUPRC: 30.74 -> 36.17 | Improved |
| add cross-domain feature alignment to ARC (ablation) | ARC · 8 test graphs average | AUPRC: 30.74 -> 31.74 | Improved |
| add cross-domain feature alignment to ARC (ablation) | ARC · Weibo | AUPRC: 64.18 -> 61.79 | Hurt |
| add cross-domain feature alignment to ARC (ablation) | ARC · 8 test graphs average | AUROC: 74.04 -> 74.61 | Improved |
| remove feature normalization, structural patterns, or truncated attention (ablation) | OwlEye · 8 test graphs average | AUPRC | Improved |
| add patterns from auxiliary graphs to dictionary without finetuning (ablation) | OwlEye · Facebook/Weibo/Reddit/Amazon average | AUPRC: 31.29 -> 32.27 | Improved |
| finetune on 3 auxiliary graphs (ablation) | OwlEye · Facebook/Weibo/Reddit/Amazon average | AUPRC: 31.29 -> 31.33 | No clear effect |
| increase dictionary size from 10 to 200 patterns (ablation) | OwlEye · 8 test graphs average | AUPRC: 35.46 -> 36.01 | Improved |
| increase dictionary size beyond 200 (ablation) | OwlEye · 8 test graphs average | AUPRC: 36.01 -> 36.17 | No clear effect |
| use 10-shot labels with OwlEye vs zero-shot | OwlEye · 8 test graphs average | AUPRC: 36.17 -> 36.73 | Improved |
Closest papers by meaning
- ARC: A Generalist Graph Anomaly Detector with In-Context Learning (NeurIPS 2024 · similarity 0.90)
- Graph Anomaly Detection as Dynamical Transport: Training-Free Scoring via Empirical Bayes (NeurIPS 2026 · similarity 0.90)
- Correcting False Alarms from Unseen: Adapting Graph Anomaly Detectors at Test Time (AAAI 2026 · similarity 0.90)
- UniGAD: Unifying Multi-level Graph Anomaly Detection (NeurIPS 2024 · similarity 0.89)
- Open-Set Graph Anomaly Detection via Normal Structure Regularisation (ICLR 2025 · similarity 0.88)
- DiffGAD: A Diffusion-based Unsupervised Graph Anomaly Detector (ICLR 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.
