Table 1 Parameters of the BiLSTM neural network model
Hyper parameters | Value |
|---|---|
Tokenizer max word features | 10,000 |
Maximum sequence length | 70 (mean average sequence length) |
Bidirectional LSTM layer cell size | 140 per layer |
Attention units | 10 |
Fully connected layer dimension | 20 |
Output layer activation | Sigmoid |
Dropout fraction | 0.05 |
Training batch size | 100 |
Training epochs | 30 with early stopping monitoring change in the validation loss value |