University of Montreal

Using Deep Learning to pre-process Electrophysiology Data

• Implemented a sequential Variational Auto Encoder (VAE) network to learn a mapping from raw EEG data to clean EEG data
• Explored loss functions like MSE, KL Divergence, Gaussian Negative Log Likelihood to capture the target data
• Performed an ablation study to understand the roles of RNN, VAE components
• Implemented EEG quality assessment metrics to quantify the goodness of the EEG signals