Research map / Graph learning / Graph self-supervised learning
LUNAR: Unifying Local Outlier Detection Methods via Graph Neural Networks
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 |
|---|---|---|---|
| replace fixed local-outlier scoring with learnable neighbour-distance aggregation | LUNAR · HRSS | AUC: 60.20 -> 92.17 | Helped |
| replace fixed local-outlier scoring with learnable neighbour-distance aggregation | LUNAR · MI-V | AUC: 84.28 -> 96.73 | Helped |
| replace fixed local-outlier scoring with learnable neighbour-distance aggregation | LUNAR · THYROID | AUC: 76.30 -> 85.44 | Helped |
| use Subspace Perturbation negatives instead of Uniform negatives (ablation) | LUNAR · most datasets | AUC | Helped |
| use Uniform negatives instead of Subspace Perturbation negatives (ablation) | LUNAR · OPTDIGITS | AUC: 93.81 -> 99.86 | Helped |
| mix Subspace Perturbation and Uniform negatives instead of SP only (ablation) | LUNAR · MI-F | AUC: 84.17 -> 84.37 | Helped |
| mix Subspace Perturbation and Uniform negatives instead of SP only (ablation) | LUNAR · MI-V | AUC: 96.64 -> 96.73 | Helped |
| mix Subspace Perturbation and Uniform negatives instead of SP only (ablation) | LUNAR · SHUTTLE | AUC: 99.96 -> 99.97 | Helped |
| mix Subspace Perturbation and Uniform negatives instead of SP only (ablation) | LUNAR · HRSS | AUC: 93.32 -> 92.17 | Hurt |
| mix Subspace Perturbation and Uniform negatives instead of SP only (ablation) | LUNAR · THYROID | AUC: 85.99 -> 85.44 | Hurt |
| mix Subspace Perturbation and Uniform negatives instead of SP only (ablation) | LUNAR · SATELLITE | AUC: 85.37 -> 85.35 | No clear effect |
| vary neighbourhood size k from 2 to 200 (ablation) | LUNAR · HRSS | AUC: 93.88 -> 90.09 | Hurt |
Closest papers by meaning
- Unifying Unsupervised Graph-Level Anomaly Detection and Out-of-Distribution Detection: A Benchmark (ICLR 2025 · similarity 0.87)
- ARC: A Generalist Graph Anomaly Detector with In-Context Learning (NeurIPS 2024 · similarity 0.86)
- Self-Discriminative Modeling for Anomalous Graph Detection (ICML 2025 · similarity 0.86)
- DiffGAD: A Diffusion-based Unsupervised Graph Anomaly Detector (ICLR 2025 · similarity 0.86)
- SpaceGNN: Multi-Space Graph Neural Network for Node Anomaly Detection with Extremely Limited Labels (ICLR 2025 · similarity 0.86)
- ALTBI: Constructing Improved Outlier Detection Models via Optimization of Inlier-Memorization Effect (AAAI 2025 · 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.