Research map / Graph learning / Graph self-supervised learning
Modeling Spectral Energy Shifts in Spatio-Temporal Graph Anomaly Detection
Yilin Liu, Hongchao Zhang, Ahmad Taha, Taylor T Johnson, Meiyi Ma
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 |
|---|---|---|---|
| add left-shift energy branch to right-shift-only energy (ablation) | EGNN · YelpChi | Macro-F1: 72.21 -> 76.89 | Improved |
| add right-shift energy branch to left-shift-only energy (ablation) | EGNN · YelpChi | Macro-F1: 71.77 -> 76.89 | Improved |
| add left-shift energy to right-shift-only energy (ablation) | EGNN · Amazon | Macro-F1: 87.20 -> 91.52 | Improved |
| use learnable MLP gate fusion instead of fixed equal-weight fusion (ablation) | EGNN · Amazon | Macro-F1: 72.86 -> 91.52 | Improved |
| use learnable MLP gate fusion instead of fixed equal-weight fusion (ablation) | EGNN · YelpChi | Macro-F1: 49.44 -> 76.89 | Improved |
| use left-shift energy instead of random-walk SRW feature (ablation) | EGNN · YelpChi | Macro-F1: 50.31 -> 76.89 | Improved |
| use GraphSAGE backbone instead of GCN (ablation) | EGNN · YelpChi | Macro-F1: 58.13 -> 76.89 | Improved |
| use GraphSAGE backbone instead of GAT (ablation) | EGNN · Amazon | Macro-F1: 60.03 -> 91.52 | Improved |
| use MLP gate instead of cross-attention gate (ablation) | EGNN · YelpChi | Macro-F1: 75.19 -> 76.89 | Improved |
| use MLP gate instead of bilinear fusion (ablation) | EGNN · Amazon | Macro-F1: 88.78 -> 91.52 | Improved |
| increase neighborhood from 1-hop to 3-hop (ablation) | EGNN · YelpChi | Macro-F1: 76.91 -> 78.21 | Improved |
| increase neighborhood from 3-hop to 7-hop (ablation) | EGNN · YelpChi | Macro-F1: 78.21 -> 74.36 | Hurt |
Closest papers by meaning
- Graph Anomaly Detection as Dynamical Transport: Training-Free Scoring via Empirical Bayes (NeurIPS 2026 · similarity 0.89)
- DiffGAD: A Diffusion-based Unsupervised Graph Anomaly Detector (ICLR 2025 · similarity 0.88)
- UniFORM: Towards Unified Framework for Anomaly Detection on Graphs (AAAI 2025 · similarity 0.88)
- Leveraging Diffusion Model as Pseudo-Anomalous Graph Generator for Graph-Level Anomaly Detection (ICML 2025 · similarity 0.88)
- Dynamic Spectral Graph Anomaly Detection (AAAI 2025 · similarity 0.88)
- ARC: A Generalist Graph Anomaly Detector with In-Context Learning (NeurIPS 2024 · 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.
