Table 5.
Accuracy of NBI-Net models with softmax and linear SVM
| Model | Softmax | Linear SVM | Classifier boosting |
|---|---|---|---|
| NBI-Net | 90.03 | 90.93 | 0.90 |
| NBI-Net Aug | 92.31 | 92.74 | 0.43 |
CNN = NBI-Net with softmax
CNN-SVM = NBI-Net with linear SVM
Aug (Augmentation) is optional for both CNN and CNN-SVM
The top results have been styled with bold and italic