Node-level Regression Uncertainty Quantification on BA
100PICPER-GNN
Evaluation results
| Method | Links | ||
|---|---|---|---|
| ER-GNN2026.05 | 100 | 6.89 | |
| BayesianNN2026.05 | 100 | 3.08 | |
| QpiGNNWidth penalty (lambda)=0.52026.05 | 98 | 0.49 | |
| QpiGNNWidth penalty (lambda)=0.12026.05 | 98 | 1.01 | |
| QpiGNNWidth penalty (lambda)=opt.2026.05 | 98 | 0.48 | |
| CF-GNN2026.05 | 90 | 68.27 | |
| SQR-GNN2026.05 | 81 | 0.72 | |
| RQRadj.-GNN2026.05 | 78 | 0.75 | |
| MC Dropout2026.05 | 67 | 0.26 |