- [ ] Restructure notebooks
- [ ] Implement proper cross attention
- [ ] Implement modern UNET architectures
- [ ] Retrain for very low loss
This is a deep learning classification of EEG signals using CNN, CNN-LSTM, CNN-MHA. Training data was also used to create a 1D diffusion model that would produce synthetic EEG samples using a linear beta scheduler.
Here is a link to the paper containing the motivation and process behind the project. This was done in a similar style to CVPR format.