Table 5. The classification accuracy of each individual feature using SVM, RF, and QDA classifiers for BreakHis dataset.
| Features | SVM | RF | QDA |
|---|---|---|---|
| Magnification Factor 40× | |||
| Inception-Resnet V2 | 96.31 | 95.72 | 93.04 |
| ResNet-50 | 95.65 | 91.33 | 93.78 |
| GoogleNet | 94.29 | 95.09 | 92.58 |
| AlexNet | 91.98 | 92.44 | 94.84 |
| ShuffleNet | 86.79 | 88.02 | 90.05 |
| HOG | 90.75 | 93.01 | 89.99 |
| WPD | 88.91 | 92.27 | 91.77 |
| Magnification Factor 100× | |||
| Inception-Resnet V2 | 95.33 | 96.83 | 93.09 |
| ResNet-50 | 93.11 | 94.71 | 95.35 |
| GoogleNet | 92.99 | 94.89 | 90.84 |
| AlexNet | 88.69 | 91.35 | 92.65 |
| ShuffleNet | 86.25 | 88.93 | 87.82 |
| HOG | 91.24 | 90.49 | 80.09 |
| WPD | 88.77 | 90.76 | 86.31 |
| Magnification Factor 200× | |||
| Inception-Resnet V2 | 92.22 | 94.94 | 90.52 |
| ResNet-50 | 90.43 | 92.60 | 93.00 |
| GoogleNet | 90.45 | 92.20 | 89.11 |
| AlexNet | 88.51 | 90.94 | 87.31 |
| ShuffleNet | 85.13 | 81.46 | 78.99 |
| HOG | 87.46 | 85.31 | 88.45 |
| WPD | 82.21 | 87.32 | 79.45 |
| Magnification Factor 400× | |||
| Inception-Resnet V2 | 92.994 | 90.122 | 87.432 |
| ResNet-50 | 89.456 | 88.321 | 90.322 |
| GoogleNet | 86.666 | 89.300 | 87.199 |
| AlexNet | 84.486 | 87.432 | 88.444 |
| ShuffleNet | 82.345 | 81.111 | 80.340 |
| HOG | 84.776 | 85.241 | 83.888 |
| WPD | 85.828 | 83.102 | 81.442 |