| Model | Architecture | Learning rate | Accuracy |
| CNN | Conv1 (, ) | 0.001 | 85.41% | Maxpooling (2, 2) | Conv2 (, ) | Maxpooling (2, 2) | Conv3 (, ) | Maxpooling (2, 2) | FC1 (256, 128) | FC2 (128, 16) |
| LSTM | LSTM(128) | 0.0001 | 64.89% | FC(128, 16) |
| CNN-LSTM | Conv1 (, ) | 0.001 | 93.36% | Maxpooling (2, 2) | Conv2 (, ) | Maxpooling (2, 2) | Conv3 (, ) | Maxpooling (2, 2) | LSTM(256) | FC1 (256, 128) | FC2 (128, 16) |
| LSTM-CNN | LSTM(128) | 0.001 | 99.68% | Conv1 (, ) | Maxpooling (2, 2) | Conv2 (, ) | Maxpooling (2, 2) | Conv3 (, ) | Maxpooling (2, 2) | FC1 (256, 128) | FC2 (128, 16) |
|
|