Semi-supervised classification on NIFD (10% train, 90% test)
72.3AccuracyARMA-C3
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| ARMA-C3Loss Function=–, λcon=0.012026.05 | 72.3 | 81.2 | 77.1 | 78.9 | |
| ARMALoss Function=MERIT + Cut, λcon=0.012026.05 | 72.3 | 81.2 | 77.3 | 78.9 | |
| ARMALoss Function=Mod [29], λcon=0.012026.05 | 71.3 | 76.4 | 84.6 | 79.6 | |
| GATLoss Function=MERIT[11] + Mod, λcon=0.0012026.05 | 71.1 | 80.6 | 75.7 | 77.8 | |
| SVCLoss Function=–, λcon=–2026.05 | 70.8 | 73 | 90.3 | 80.6 | |
| GATLoss Function=MERIT + Cut, λcon=0.00012026.05 | 70.3 | 80 | 75.3 | 77.3 | |
| Random ForestLoss Function=–, λcon=–2026.05 | 70.2 | 72.9 | 89.4 | 80 | |
| GATLoss Function=Mod, λcon=0.52026.05 | 70 | 74.7 | 85.8 | 79.1 | |
| JacobiConvLoss Function=–, λcon=–2026.05 | 69.2 | 77.8 | 76.9 | 76.9 | |
| APPNPLoss Function=–, λcon=–2026.05 | 69 | 78.1 | 75.7 | 76.6 | |
| GATLoss Function=Cut, λcon=0.52026.05 | 68.2 | 85.6 | 63.5 | 72.9 | |
| MLPClassifierLoss Function=–, λcon=–2026.05 | 68.1 | 79 | 72.3 | 75.3 | |
| ARMALoss Function=DGI[27] + Mod, λcon=0.0092026.05 | 68.1 | 77.6 | 75.1 | 75.5 | |
| ARMALoss Function=Cut[48], λcon=0.92026.05 | 67.5 | 84.7 | 63.4 | 72.5 | |
| ARMALoss Function=DGI + Cut, λcon=0.0012026.05 | 67.3 | 77 | 75.5 | 74.4 | |
| GATLoss Function=DGI + Mod, λcon=0.0012026.05 | 65.6 | 79.3 | 65.9 | 71.8 | |
| GATLoss Function=Baseline, λcon=–2026.05 | 63.9 | 80.6 | 62.1 | 69.9 | |
| GATLoss Function=DGI + Cut, λcon=0.0012026.05 | 63.7 | 81.9 | 58.7 | 68.4 | |
| ARMALoss Function=Baseline, λcon=–2026.05 | 63 | 78.5 | 62.5 | 69.5 | |
| XGBoostLoss Function=–, λcon=–2026.05 | 60.4 | 70.3 | 71.1 | 69.4 |