Node-level Regression Uncertainty Quantification on Outlier
100PICPBayesianNN
Evaluation results
| Method | Links | ||
|---|---|---|---|
| BayesianNN2026.05 | 100 | 2.95 | |
| QpiGNNWidth penalty (lambda)=0.12026.05 | 99 | 0.75 | |
| ER-GNN2026.05 | 97 | 0.88 | |
| CF-GNN2026.05 | 93 | 1.92 | |
| SQR-GNN2026.05 | 90 | 0.1 | |
| RQRadj.-GNN2026.05 | 90 | 0.36 | |
| QpiGNNWidth penalty (lambda)=opt.2026.05 | 90 | 0.49 | |
| QpiGNNWidth penalty (lambda)=0.52026.05 | 89 | 0.47 | |
| MC Dropout2026.05 | 58 | 0.06 |