Research map / Graph learning / Graph self-supervised learning
Learnable Kernel Density Estimation for Graphs and Its Application to Graph-Level Anomaly Detection
Xudong Wang, Ziheng Sun, Chris Ding, Jicong Fan
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 multi-scale KDE instead of the best single bandwidth h=10^-1 (ablation) | LGKDE · MUTAG | AUROC: 86.92 -> 91.63 | Improved |
| use bandwidth 10^-1 rather than 10^2 in single-scale KDE (ablation) | LGKDE · MUTAG | AUROC: 80.89 -> 86.92 | Improved |
| add MMD distance instead of average graph readout (ablation) | LGKDE · MUTAG | AUROC: 83.89 -> 91.63 | Improved |
| add MMD distance instead of sum graph readout (ablation) | LGKDE · MUTAG | AUROC: 84.73 -> 91.63 | Improved |
| add learnable mixture weights instead of fixed weights (ablation) | LGKDE · MUTAG | AUROC: 88.92 -> 91.63 | Improved |
| add learnable mixture weights (ablation) | LGKDE · MUTAG | FPR95: 36.97 -> 30.80 | Improved |
| add multi-scale KDE instead of h=10^-1 (ablation) | LGKDE · MUTAG | FPR95: 37.95 -> 30.80 | Improved |
| jointly train metric and density end-to-end instead of GAE then separate KDE | LGKDE · MUTAG | AUROC: 81.94 -> 91.63 | Improved |
| use energy-based spectral perturbation versus random perturbations (ablation) | LGKDE · graph anomaly detection | AUROC | Improved |
| use moderate GNN depth of 2-3 layers (ablation) | LGKDE · MUTAG | AUROC | Improved |
| LGKDE versus best baseline SIGNET on DHFR | LGKDE · DHFR | AUROC: 72.87 -> 82.58 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- UniFORM — UniFORM: Towards Unified Framework for Anomaly Detection on Graphs (AAAI 2025) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- DiffGAD: A Diffusion-based Unsupervised Graph Anomaly Detector (ICLR 2025 · similarity 0.87)
- Graph Anomaly Detection as Dynamical Transport: Training-Free Scoring via Empirical Bayes (NeurIPS 2026 · similarity 0.87)
- General Graph Random Features (ICLR 2024 · similarity 0.87)
- Dynamic Similarity Graph Construction with Kernel Density Estimation (ICML 2025 · similarity 0.87)
- Leveraging Diffusion Model as Pseudo-Anomalous Graph Generator for Graph-Level Anomaly Detection (ICML 2025 · similarity 0.86)
- Unifying Unsupervised Graph-Level Anomaly Detection and Out-of-Distribution Detection: A Benchmark (ICLR 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.
