Node-level Regression Uncertainty Quantification on Gaussian
100PICPER-GNN
Evaluation results
| Method | Links | ||
|---|---|---|---|
| ER-GNN2026.05 | 100 | 1.85 | |
| BayesianNN2026.05 | 100 | 2.98 | |
| QpiGNNWidth penalty (lambda)=0.12026.05 | 99 | 0.84 | |
| QpiGNNWidth penalty (lambda)=opt.2026.05 | 95 | 0.64 | |
| QpiGNNWidth penalty (lambda)=0.52026.05 | 92 | 0.55 | |
| CF-GNN2026.05 | 91 | 2.9 | |
| SQR-GNN2026.05 | 88 | 0.5 | |
| RQRadj.-GNN2026.05 | 88 | 0.53 | |
| MC Dropout2026.05 | 55 | 0.2 |