Node-level Regression Uncertainty Quantification on Income
100PICPER-GNN
Evaluation results
| Method | Links | ||
|---|---|---|---|
| ER-GNN2026.05 | 100 | 3.12 | |
| BayesianNN2026.05 | 100 | 2.97 | |
| QpiGNNWidth penalty (lambda)=0.12026.05 | 100 | 0.72 | |
| QpiGNNWidth penalty (lambda)=0.52026.05 | 99 | 0.41 | |
| QpiGNNWidth penalty (lambda)=opt.2026.05 | 99 | 0.44 | |
| CF-GNN2026.05 | 92 | 3.48 | |
| CF-GNNoptimized=true2026.05 | 91 | 2.92 | |
| RQRadj.-GNN2026.05 | 89 | 0.36 | |
| SQR-GNN2026.05 | 86 | 0.22 | |
| MC Dropout2026.05 | 45 | 0.09 |