Table 4.
Standard deviation of classification accuracies for each network model.
| Model’s name/dataset | Kaggle Cats and Dogs | CIFAR-10 | MNIST |
|---|---|---|---|
| AlexNet (horizontally shifted test images) | – | – | 9.98 |
| AlexNet (vertically shifted test images) | – | – | 16.59 |
| Trans-ONN-vert | 2.49 | 7.18 | 15.36 |
| Trans-ONN-vert-GAP | 0.7 | 4.66 | 0.91 |
| Trans-ONN-horiz | 3.29 | 7.92 | 19.44 |
| Trans-ONN-horiz-GAP | 1.12 | 4.51 | 5.69 |
| Trans-ONN-cascaded | 3.28 | 8.29 | 17.22 |
| Trans-ONN-cascaded-GAP | 2.25 | 5.51 | 10.99 |