This repository contains our work done for ECE 239 AS, where we used deep learning models including RNNs, VAEs, and CNNs in order to obtain high classification accuracy on patients' actions given EEG data. The dataset consited of 255 trials, where each trial had 22 electrode values across 1000 timesteps. In addition, we had access to data for nine patients, each of whom had 255 individual trials.
Built by Rohan Varma Jennifer Liaw and Michael Moon
- Download the data and place it in a directory called
project_datasets/
- The code requires Python3 & Pytorch to run. Each file has command line arguments that can be used to run the code.