| Task | Description | Team Member | Internal Deadline | Completed? |
|---|---|---|---|---|
| Data preprocessing | Retrieve and label audio clip datasets | Paul, Saad | Feb 18 | Y |
| Audio Feature Expansion | Use librosa library to extract features | Paul, Saad | Feb 22 | Y |
| Baseline Model | Develop random forest model and perform end-to-end testing | Muhammad, Saad | Feb 25 | Y |
| Initial NN Model | Build Neural Network Model using NLS + feedforward ANN | Muhammad, Mahmoud | Feb 27 | Y |
| Progress Report | Work on progress report | ALL | Mar 02 | In Progress |
| LSTM Model | Modify initial NN to include LSTM layers | Muhammad, Mahmoud | Mar 07 | |
| Tune Hyperparameters (learning) | Investigate and analyze effect of learning rate, batch size, dropout rate | Paul, Saad | Mar 14 | |
| Tune Hyperparameters (LSTM) | Investigate different LSTM layers and Uni vs Bidirectional | Muhammad, Mahmoud | Mar 20 | |
| Finalize Model | Based on all gathered findings, finalize the model before proceeding with test data | ALL | Mar 30 | |
| Project Presentation | Prepare slides and demo | ALL | April 06 | |
| Final Deliverable | Prepare final report and summarize findings | ALL | April 13 |
| Test Accuracy | |
|---|---|
| Baseline | 53.53% |
| Initial NN | 64.11% |
NSL with keras: tensorflow NSL
NSL tutorial: tutorial
Cuda with tensorflow: tf cuda