Research map / Graph learning / Graph self-supervised learning
Healthcare Insurance Fraud Detection via Continual Fiedler Vector Graph Model
Yehan Zhang, Huaidong Zhang, Xuandi Luo, Shengfeng 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 |
|---|---|---|---|
| add Fiedler-vector-guided graph autoencoder (pretraining) (ablation) | ConFVG · Medical insurance (10% label rate) | AUC: 67.21 -> 76.13 | Improved |
| add subgraph graph complement on top of autoencoder (ablation) | ConFVG · Medical insurance (10% label rate) | AUC: 76.13 -> 78.21 | Improved |
| add mean-teacher on top of autoencoder (ablation) | ConFVG · Medical insurance (10% label rate) | AUC: 76.13 -> 77.35 | Improved |
| add mean-teacher to graph complement without autoencoder (ablation) | ConFVG · Medical insurance (10% label rate) | AUC: 68.35 -> 72.08 | Improved |
| add mean-teacher alone to a model with no other component (ablation) | ConFVG · Medical insurance (10% label rate) | AUC: 67.21 -> 67.43 | No clear effect |
| add all three components vs autoencoder + graph complement (ablation) | ConFVG · Medical insurance (10% label rate) | AUC: 78.21 -> 80.48 | Improved |
| use Fiedler-vector mask instead of random mask (ablation) | ConFVG · Medical insurance (10% label rate) | AUC: 74.14 -> 80.48 | Improved |
| use Fiedler-vector mask instead of VGAE (ablation) | ConFVG · Medical insurance (10% label rate) | AUC: 77.87 -> 80.48 | Improved |
| raise top-k component ratio nc from 0.05 to 0.3 (ablation) | ConFVG · Medical insurance (10% label rate) | AUC: 80.48 -> 78.08 | Hurt |
| raise perturbation epsilon from 1e-5 to 1e-2 (ablation) | ConFVG · Medical insurance (10% label rate) | AUC: 80.48 -> 76.05 | Hurt |
| raise attention loss weight alpha_attn from 0.1 to 2 (ablation) | ConFVG · Medical insurance (10% label rate) | AUC: 80.48 -> 76.22 | Hurt |
| raise EMA decay from 0.99 to 0.999 (ablation) | ConFVG · Medical insurance (10% label rate) | AUC: 80.48 -> 78.82 | Hurt |
Closest papers by meaning
- Dynamic Neighborhood Modeling via Node-Subgraph Contrastive Learning for Graph-Based Fraud Detection (AAAI 2025 · similarity 0.87)
- Barely Supervised Learning for Graph-Based Fraud Detection (AAAI 2024 · similarity 0.86)
- Context-aware Graph Neural Network for Graph-based Fraud Detection with Extremely Limited Labels (AAAI 2025 · similarity 0.86)
- Pre-trained Online Contrastive Learning for Insurance Fraud Detection (AAAI 2024 · similarity 0.86)
- A Label-free Heterophily-guided Approach for Unsupervised Graph Fraud Detection (AAAI 2025 · similarity 0.86)
- Partitioning Message Passing for Graph Fraud Detection (ICLR 2024 · similarity 0.85)
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.
