Copyright (C) 2020 ETH Zurich, Switzerland. SPDX-License-Identifier: Apache-2.0. See LICENSE file for details.

Authors: Xiaying Wang, Michael Hersche

In this repository, we share the code for classifying MI data of the Physionet EEG Motor Movement/Imagery Dataset using EEGNet. For details, please refer to the papers below.

If this code proves useful for your research, please cite

Xiaying Wang, Michael Hersche, Batuhan Tömekce, Burak Kaya, Michele Magno, Luca Benini, "An Accurate EEGNet-based Motor-Imagery Brain--Computer Interface for Low-Power Edge Computing", in IEEE International Symposium on Medical Measurements and Applications (MEMEA), 2020.

You will need a machine with a CUDA-enabled GPU and the Nvidia SDK installed to compile the CUDA kernels. Further, we have used conda as a python package manager and exported the environment specifications to dependency.yml. You can recreate our environment by running

conda env create -f dependency.yml -n mybciEnv 

Make sure to activate the environment before running any code.

If get a batchnormalization error when running the code, follow the instructions by gurjar112 here:

change "()" to "[ ]" in line no 1908,1910,1914, 1918 in anaconda3/envs/mybciEnv/lib/python3.6/site-packages/keras/backend/tensorflow_backend.py (e.g. beta = tf.reshape(beta, [-1])) and have 'channels_last' in keras.json

EEGNet: Download the .edf files of the Physionet EEG Motor Movement/Imagery Dataset here, unpack it, and put into the folder dataset/

Global models are trained and validated in main_global.py. Results are in results/your-global-experiment/stats and global models in results/your-experiment-name/model.

$ (mybciEnv) python3 main_global.py

After having trained and stored the global model, they can be refined by doing subject-specific transfer learning (SS-TL) using main_ss.py.

$ (mybciEnv) python3 main_ss.py

A notebook with selected plots is available under plot_nbook/.

Please refer to the LICENSE file for the licensing of our code.