Table 1 Accuracies of the classifiers composing the ensemble
From: Model order reduction assisted by deep neural networks (ROM-net)
Model | Type | Loss function | Test accuracy (\(\%\)) |
|---|---|---|---|
1 | FC | Cross-entropy | 72.50 |
2 | FC | Balanced cross-entropy | 71.90 |
3 | FC | Focal loss | 67.95 |
4 | FC | Focal loss | 68.45 |
5 | CNN | Cross-entropy | 72.05 |
6 | CNN | Cross-entropy | 72.45 |
7 | CNN | Cross-entropy | 73.75 |
8 | CNN | Balanced cross-entropy | 72.00 |
9 | CNN | Focal loss | 69.45 |
10 | GAP-CNN | Cross-entropy | 63.05 |
11 | GAP-CNN | Cross-entropy | 68.00 |
12 | GAP-CNN | Balanced cross-entropy | 70.05 |
Ensemble | Unweighted averaging | – | 80.00 |