Table 4.
Hyperparameters of all models used to test the focal loss function presented in Section 3.2.
| Deep Learning Architecture | Hyperparameters |
|---|---|
| Bi-LSTM | Number of layers: 2 |
| Layer 1 units: 32 Layer 2 units: 16 |
|
| Activation function: ReLU | |
| DNN | Number of layers: 2 |
| Layer 1 units: 64 Layer 2 units: 64 |
|
| Activation function: Sigmoid | |
| CNN1D | Number of layers: 1 |
| Layer 1 units: 30 | |
| Activation function: ReLU | |
| Filter size: 10 × 1 | |
| Bi-GRU | Number of layers: 2 |
| Layer 1 units: 32 Layer 2 units: 16 |
|
| Activation function: ReLU |