Node-level Regression Uncertainty Quantification on Education
100PICPER-GNN
Evaluation results
| Method | Links | ||
|---|---|---|---|
| ER-GNN2026.05 | 100 | 4.37 | |
| BayesianNN2026.05 | 100 | 2.96 | |
| QpiGNNWidth penalty (lambda)=0.52026.05 | 99 | 0.57 | |
| QpiGNNWidth penalty (lambda)=0.12026.05 | 99 | 0.9 | |
| QpiGNNWidth penalty (lambda)=opt.2026.05 | 99 | 0.59 | |
| CF-GNNoptimized=true2026.05 | 90 | 3.1 | |
| SQR-GNN2026.05 | 88 | 0.32 | |
| CF-GNN2026.05 | 88 | 2.78 | |
| RQRadj.-GNN2026.05 | 87 | 0.49 | |
| MC Dropout2026.05 | 40 | 0.11 |