Node-level Regression Uncertainty Quantification on Squirrel
100PICPBayesianNN
Evaluation results
| Method | Links | ||
|---|---|---|---|
| BayesianNN2026.05 | 100 | 2.97 | |
| QpiGNNWidth penalty (lambda)=0.12026.05 | 99 | 0.47 | |
| ER-GNN2026.05 | 97 | 0.8 | |
| QpiGNNWidth penalty (lambda)=opt.2026.05 | 96 | 0.18 | |
| RQRadj.-GNN2026.05 | 89 | 0.15 | |
| QpiGNNWidth penalty (lambda)=0.52026.05 | 73 | 0.07 | |
| MC Dropout2026.05 | 31 | 0.02 | |
| SQR-GNN2026.05 | 22 | 0.01 |