Research map / Graph learning / Graph self-supervised learning
Graph Anomaly Detection via Multi-Scale Contrastive Learning Networks with Augmented View
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 subgraph-subgraph contrast to node-subgraph contrast, without node-node contrast (ablation) | GRADATE · EAT, WebKB, UAT, Cora, UAI2010, Citation | AUC: 0.6762 -> 0.6800; 0.7949 -> 0.8310; 0.6538 -> 0.6603; 0.8847 -> 0.8956; 0.8175 -> 0.9055; 0.7296 -> 0.6978 | Mixed |
| Add subgraph-subgraph contrast to node-subgraph plus node-node contrast (ablation) | GRADATE · EAT, WebKB, UAT, Cora, UAI2010, Citation | AUC: 0.7726 -> 0.7980; 0.8208 -> 0.8740; 0.8087 -> 0.8451; 0.9122 -> 0.9237; 0.8731 -> 0.9262; 0.8028 -> 0.8138 | Helped |
| Use node-subgraph contrast only instead of all three contrasts (ablation) | GRADATE · EAT, WebKB, UAT, Cora, UAI2010, Citation | AUC: 0.7980 -> 0.6762; 0.8740 -> 0.7949; 0.8451 -> 0.6538; 0.9237 -> 0.8847; 0.9262 -> 0.8175; 0.8138 -> 0.7296 | Hurt |
| Replace edge modification with Gaussian noised features, feature masking, or graph diffusion (ablation) | GRADATE · EAT, WebKB, UAT, Cora, UAI2010, Citation | AUC | Hurt |
| Perturb node features for the augmented view (ablation) | GRADATE · Six benchmark datasets | AUC | Hurt |
| Use graph diffusion rather than edge modification for augmentation (ablation) | GRADATE · Six benchmark datasets | AUC | Hurt |
| Set the subgraph-subgraph loss weight γ to 0.1 (ablation) | GRADATE · All benchmarks | AUC | Helped |
| Vary edge modification proportion P (ablation) | GRADATE · UAT, UAI2010, Citation | AUC | No clear effect |
Closest papers by meaning
- Graph Anomaly Detection with Diffusion Model-Based Graph Enhancement (Student Abstract) (AAAI 2024 · similarity 0.91)
- ARC: A Generalist Graph Anomaly Detector with In-Context Learning (NeurIPS 2024 · similarity 0.89)
- Open-Set Graph Anomaly Detection via Normal Structure Regularisation (ICLR 2025 · similarity 0.89)
- Revisiting Graph Contrastive Learning on Anomaly Detection: A Structural Imbalance Perspective (AAAI 2025 · similarity 0.88)
- UniGAD: Unifying Multi-level Graph Anomaly Detection (NeurIPS 2024 · similarity 0.88)
- Disentangling Invariant Subgraph via Variance Contrastive Estimation under Distribution Shifts (ICML 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.