Research Article
Improved Classification of White Blood Cells with the Generative Adversarial Network and Deep Convolutional Neural Network
Table 10
10-fold cross-validation accuracy of the DNN models initialized using pretrained weights.
| Model | Tran_aug1 + GAN_aug1 (200 inst./class) (%) | Tran_aug2 + GAN_aug2 (300 inst./class) (%) | Tran_aug3 + GAN_aug3 (400 inst./class) (%) |
| VGG-16 | 94.3 | 94.9 | 95.7 | VGG-19 | 94.8 | 95.4 | 95.9 | ResNet-18 | 94.1 | 95.2 | 95.4 | ResNet-50 | 95.8 | 96.7 | 97.4 | DenseNet-121 | 96.3 | 97.4 | 98.3 | DenseNet-169 | 96.9 | 98.1 | 98.8 |
|
|