This task is centered around the analysis of EEG data to classify different cognitive states using advanced deep learning techniques.
- EEGNet as it was specifically designed for EEG signal classification. It is lightweight, making it suitable for real-time applications and deployment on devices with limited computational power.
- TSCeption as it can handle the temporal and spectral characteristics of EEG signals
- ATCNet because it is designed to handle temporal dependencies in EEG data.
The rest state shows a higher power across all frequencies than the task state. These results seem to support notions I read that brain activity is better during rest/sleep. Be it retention, dreams, healing etc. more
| Model | Accuracy | Precision | Recall | F1-Score |
|---|---|---|---|---|
| EEGNet | 0.8000 | 0.7000 | 1.0000 | 0.8235 |
| TSCeption | 0.8000 | 0.8100 | 0.8000 | 0.8000 |
| ATCNet | 0.4667 | 0.4667 | 1.0000 | 0.6364 |
While fine-tuning the hyperparameters and experimenting with the architecture, preference was given to Higher Recall. Because we are dealing with EEG signals. In cases like medical diagnostics missing a positive case is critical.