/BCI_EEGNet

EEGNet implementation of 2 BCI competition datasets

Primary LanguageJupyter Notebook

BCI_EEGNet

EEGNet implementation of 2 BCI competition datasets:

  1. Kaggle competition Dataset: https://www.kaggle.com/c/inria-bci-challenge/data

  2. BCI Competition III Dataset 2: http://www.bbci.de/competition/iii/#data_set_ii

EEGNet CNN architecture PyTorch implementation borrowed from : Sriram Ravindran: https://github.com/aliasvishnu/EEGNet

All the data used in the codes was earlier bandpassed filtered in MATLAB with a 2nd order Butterworth Filter from 0.1-30 Hz

FLOW:

  1. Run the .m filtering file on the dataset obtained from the link for the BCI COmpetition Dataset

  2. Run the file BCI_III_DS_2_TestSet_PreProcessing.ipynb on the filtered datasets obtained from the Matlab code.

  3. RUn the BCI_III_DS_2_Filtered_Downsampled.ipynb to get results on downsampled data at 120 Hz

  4. Modify the BCI_III_DS_2_TestSet_PreProcessing.ipynb to get results at original data of 240 Hz and then run BCI_III_DS_2_Filtered Data.ipynb to get results.

  1. Run the matlab preprocessing file same as above.
  2. Run Kaggle Dataset- ERN with appropriate file path to get results.