Table 1 Hyperparameters of the CNN.
From: Contrast phase recognition in liver computer tomography using deep learning
Hyperparameter | ||
|---|---|---|
Class | Name | Possible values |
Architecture | Convolutional layers | 2, 3, 4, 5, 6 |
Architecture | Convolutional kernel | 3 \(\times\) 3, 5 \(\times\) 5, 7 \(\times\) 7 |
Architecture | Filters on 1st layer | 8, 16 |
Architecture | Dense layer size | 8, 16, 24, 32 |
Architecture | Image side | 256, 512 |
Regularizer | Dropout | [1e−6, 1e−2] |
Regularizer | Augmentation | Yes or no |
Regularizer | Batch normalization | Yes or no |
Optimizer | Optimizer | Adam, SGD |
Optimizer | Learning rate | [1e−6, 1e−2] |
