Skip to main content
. 2024 Nov 26;10(23):e40608. doi: 10.1016/j.heliyon.2024.e40608

Table 2.

Detail of U-Net training parameters.

Parameter Value
Learning rate 0.0001
Optimizer Adam
Function loss BCE
Data augmentation Vertical and horizontal flip, brightness, saturation, contrast and blurring adjustments
Maximum training iterations 100
Best performace iteration 33
Training/validation samples (HAM10000 dataset) 2340
Testing samples 1 (HAM10000 dataset) 987
Testing samples 2 (UdeC dataset) 144