Node-level Regression Uncertainty Quantification on Crocodile
100PICPER-GNN
Evaluation results
| Method | Links | ||
|---|---|---|---|
| ER-GNN2026.05 | 100 | 1.56 | |
| BayesianNN2026.05 | 100 | 3.07 | |
| QpiGNNWidth penalty (lambda)=0.12026.05 | 100 | 0.54 | |
| QpiGNNWidth penalty (lambda)=opt.2026.05 | 97 | 0.16 | |
| QpiGNNWidth penalty (lambda)=0.52026.05 | 92 | 0.08 | |
| RQRadj.-GNN2026.05 | 87 | 0.08 | |
| MC Dropout2026.05 | 46 | 0.01 | |
| SQR-GNN2026.05 | 44 | 0.01 |