Table 2 Best hyperparameters after tuning.
From: Contrast phase recognition in liver computer tomography using deep learning
Hyperparameter | ||
|---|---|---|
Class | Name | Value |
Architecture | Convolutional layers | 5 |
Architecture | Convolutional kernel | 5 \(\times\) 5 |
Architecture | Filters on 1st layer | 32 |
Architecture | Dense layer size | 32 |
Architecture | Image side | 256 |
Regularizer | L2 * | 0.01 |
Regularizer | Dropout | \(2.045 \times 10^{-4}\) |
Regularizer | Augmentation | No |
Regularizer | Batch normalization | Yes |
Optimizer | Optimizer | Adam |
Optimizer | Learning rate | 1.483 \(10^{-3}\) |
Optimizer | Loss function * | Cross entropy |
Execution | Learning epochs * | 50 |
Execution | Batch * | 5 |
