Table 2 Hyperparameter tuning of models
Model | Hyperparameters | ||
|---|---|---|---|
Layer | Batch Size | Learning rate | |
1 | Convolution 8 × 3, strides = 1 + ReLU - LSTM | 8 | 10–5 |
2 | Convolution 8 × 3, 16 × 3, strides = 1 + ReLU - LSTM | ||
3 | Convolution 8 × 3, 16 × 3, 32 × 3, strides = 1 + ReLU - LSTM | ||
4 | Convolution 8 × 3, 16 × 3, 32 × 3, 64 × 3, strides = 1 + ReLU - LSTM | ||
5 | Convolution 8 × 3, strides = 1 + ReLU - BiLSTM | ||
6 | Convolution 8 × 3, 16 × 3, strides = 1 + ReLU - BiLSTM | ||
7 | Convolution 8 × 3, 16 × 3, 32 × 3, strides = 1 + ReLU - BiLSTM | ||
8 | Convolution 8 × 3, 16 × 3, 32 × 3, 64 × 3, strides = 1 + ReLU - BiLSTM | ||
9 | Convolution 8 × 3, 16 × 3, 32 × 3, 64 × 3, 128 × 3, strides = 1 + ReLU - BiLSTM | ||
10 | Convolution 8 × 3, 16 × 3, 32 × 3, 64 × 3, 128 × 3, 256 × 3, strides = 1 + ReLU - BiLSTM | ||
11 | Convolution 8 × 3, 16 × 3, 32 × 3, 64 × 3, 128 × 3, 256 × 3, 512 × 3, strides = 1 + ReLU - BiLSTM | ||
12 | Convolution 8 × 3, 16 × 3, 32 × 3, 64 × 3, 128 × 3, 256 × 3, 512 × 3, 1024 × 3, strides = 1 + ReLU - BiLSTM | ||
13 | Convolution 8 × 3, 16 × 3, 32 × 3, 64 × 3, 128 × 3, 256 × 3, 512 × 3, 1024 × 3, 2048 × 3, strides = 1 + ReLU - BiLSTM | ||