Table 2 Values for hyperparameters investigated for training CNN models for SM-induced corneal pathology classification.
From: Artificial intelligence derived grading of mustard gas induced corneal injury and opacity
Parameter | Values |
|---|---|
# convolutional blocks | 1–5 |
# filters | 32, 64, 128 |
Kernel size | 1*1, 2*2, 3*3, 4*4 |
# dense layers | 1–5 |
# neurons | 10-1000 |
Learning rate | 2 × 10− 4, 1 × 10− 4, 2 × 10− 3, 1 × 10− 3, 1 × 10− 2, 2 × 10− 2 |
Dropout ration | 0.3, 0.5, 0.7 |
# epochs | 50, 100 |
Optimizer | Adam, Stochastic gradient descent |
