Skip to content

Graph Convolutional Networks for 4-class EEG Classification

Notifications You must be signed in to change notification settings

magnumical/GCN_for_EEG

Repository files navigation

GCN_for_EEG

Graph Convolutional Networks for 4-class EEG Classification

The Pure PYTHON interpertation!

Inspirations :

Name Description
Michaël Defferrard Created the basics of GCN!
Shuyue Jia Created awsome codes for EEG classification

What I have done?

Shuyue's work was awsome but preprocessing should be done in MATLAb, which is not available for everyone. So I interperted python version! I also added other types of GCNs to code + changed some parts of code

How to run?

  1. Download - PhysioNet 4-class EEG and place it in 01loadData folder ( or easily run downloaddata.py
  2. Run edfread.py . This code will end up in 64 electrode data + 64 Label data. USE PYTHON 2.7
  3. It's time to copy the results of previous step into 02Preprocess . I placed both MATLAB and PYTHON , But my main intention is having PURE PYTHON environment So go into WithPython and create a folder called data and place 128 .mat files there. Then run the Code It results is available in folder pythondata as .csv files
  4. create files folder where the onEEGcode.py is. Then copy CSV files there. Run the onEEGcode.py and ENJOY!

Dependencies

  1. Tensorflow 1.13
  2. Numpy
  3. Scipy
  4. Pandas
  5. Pyedflib

Some Results of Preprocessing

  1. Absolute Pearso matrix

Classification with basic GCN

Releases

No releases published

Packages

No packages published