Node-level Regression Uncertainty Quantification on Edge
100PICPER-GNN
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| ER-GNN2026.05 | 100 | 1.07 | |
| BayesianNN2026.05 | 100 | 3.06 | |
| MC Dropout2026.05 | 100 | 0.3 | |
| QpiGNNWidth penalty (lambda)=0.12026.05 | 100 | 0.97 | |
| QpiGNNWidth penalty (lambda)=0.52026.05 | 94 | 0.39 | |
| QpiGNNWidth penalty (lambda)=opt.2026.05 | 94 | 0.54 | |
| RQRadj.-GNN2026.05 | 93 | 0.83 | |
| CF-GNN2026.05 | 92 | 1.78 | |
| SQR-GNN2026.05 | 91 | 0.32 |