Table 2 CNN hyperparameters. Parameters that were used for training/testing of the CNN. Bolded parameters represents those that were used to train the final model
Hyper-parameters | Specifications |
|---|---|
Batch Size | 1 |
Optimiser | SGD, Adam |
Learning Rate | StepLR, CyclicLR, ReduceLROnPlaeau |
Loss | MSELoss, CrossEntropyLoss, BCELoss, NLLLoss |
Epochs | 50, 75, 100, 150, 250, 300, 600, 900, 1300, 1500, 1800, 2000, 2200, 2500 |