Table 1.
The shapes of outputs and the numbers of parameters in the Enhancer-LSTMAtt.
| Layers | Shape of Output | Number of Parameters |
|---|---|---|
| Input | (None, 200) | 0 |
| Embedding | (None, 200, 32) | 160 |
| Conv1D(32, 8) | (None, 200, 32) | 8224 |
| Batch Normalization | (None, 200, 32) | 128 |
| Activation | (None, 200, 32) | 0 |
| Max Pooling | (None, 199, 32) | 0 |
| Conv1D_1(64, 8) | (None, 96, 64) | 16,448 |
| Batch Normalization | (None, 96, 64) | 256 |
| Activation | (None, 96, 64) | 0 |
| Conv1D_2(64, 8) | (None, 96, 64) | 32,832 |
| Batch Normalization | (None, 96, 64) | 256 |
| Conv1D_3(64, 8) | (None, 96, 64) | 16,448 |
| Add | (None, 96, 64) | 0 |
| Activation | (None, 96, 64) | 0 |
| Conv1D_4(64, 8) | (None, 96, 64) | 32,832 |
| Batch Normalization | (None, 96, 64) | 256 |
| Activation | (None, 96, 64) | 0 |
| Conv1D_5(64, 8) | (None, 96, 64) | 32,832 |
| Batch Normalization | (None, 96, 64) | 256 |
| Add | (None, 96, 64) | 0 |
| Activation | (None, 96, 64) | 0 |
| Conv1D_6(64, 8) | (None, 96, 64) | 32,832 |
| Batch Normalization | (None, 96, 64) | 256 |
| Activation | (None, 96, 64) | 0 |
| Conv1D_7(64, 8) | (None, 96, 64) | 32,832 |
| Batch Normalization | (None, 96, 64) | 256 |
| Add | (None, 96, 64) | 0 |
| Activation | (None, 96, 64) | 0 |
| Global Max Pooling | (None, 64) | 0 |
| Bidirectional LSTM | (None, 200, 64) | 16,640 |
| Bidirectional LSTM | (None, 200, 64) | 24,832 |
| Attention | (None, 64) | 264 |
| Dropout | (None, 64) | 0 |
| Concatenate | (None, 128) | 0 |
| Dense(16) | (None, 16) | 2064 |
| Dense(1) | (None, 1) | 17 |