Table 4 Results of selected seventeen types of models/optimizers that have been selected based on average accuracy
From: Classification of mastoid air cells by CT scan images using deep learning method
Name of the architecture with its optimizer | Results in the quarter dataset based on Table 2 (%) | Results in the whole dataset | ||||
|---|---|---|---|---|---|---|
The highest accuracy (%) | Accuracy of the last epoch (%) | Average accuracy (%) | Average time of each epoch (Seconds) | |||
1 | InceptionV3_Adamax | 85.58 | 87.54 | 85.99 | 86.33 | 2928 |
2 | Xception_AdaDelta | 85.30 | 88.93 | 86.69 | 87.58 | 4588 |
3 | Xception_AdaMax | 85.13 | 88.65 | 87.80 | 87.70 | 4399 |
4 | InceptionResNetV2_Adadelta | 84.37 | 87.86 | 86.01 | 85.73 | 6833 |
5 | DenseNet201_SGD | 84.22 | 87.38 | 85.52 | 85.78 | 7292 |
6 | Xception_SGD | 83.24 | 87.64 | 86.75 | 86.75 | 4230 |
7 | Xception_Adam | 82.68 | 87.68 | 86.01 | 86.42 | 4489 |
8 | InceptionV3_Adadelta | 82.53 | 87.34 | 86.73 | 85.81 | 3135 |
9 | DenseNet121_SGD | 82.48 | 88.15 | 85.58 | 86.14 | 4594 |
10 | InceptionResNetV2_SGD | 82.45 | 86.87 | 85.75 | 85.83 | 5997 |
11 | InceptionResNetV2_Adamax | 82.42 | 87.68 | 84.76 | 85.77 | 6308 |
12 | DenseNet169_SGD | 82.42 | 88.08 | 86.41 | 86.72 | 5926 |
13 | Xception_Adagrad | 82.31 | 88.81 | 87.38 | 87.32 | 4345 |
14 | MobileNet_Adamax | 82.24 | 85.99 | 84.50 | 84.55 | 1930 |
15 | Xception_RMSprop | 82.16 | 87.80 | 85.93 | 85.19 | 4399 |
16 | InceptionResNetV2_Adagrad | 82.14 | 88.02 | 86.59 | 86.13 | 6284 |
17 | InceptionResNetV2_Adam | 82.05 | 87.84 | 84.62 | 85.96 | 6599 |