Node-level Regression Uncertainty Quantification on Uniform
100PICPER-GNN
Evaluation results
| Method | Links | ||
|---|---|---|---|
| ER-GNN2026.05 | 100 | 1.53 | |
| BayesianNN2026.05 | 100 | 3 | |
| QpiGNNWidth penalty (lambda)=0.12026.05 | 99 | 0.89 | |
| QpiGNNWidth penalty (lambda)=opt.2026.05 | 93 | 0.62 | |
| RQRadj.-GNN2026.05 | 90 | 0.68 | |
| CF-GNN2026.05 | 90 | 3.04 | |
| SQR-GNN2026.05 | 88 | 0.51 | |
| QpiGNNWidth penalty (lambda)=0.52026.05 | 88 | 0.43 | |
| MC Dropout2026.05 | 65 | 0.26 |