TSGL-EEGConv

This is the Repository of TSGL-EEGNet. TSGL-EEGNet is a kind of Convolutional Network for pre-processed EEG signals to predict classes. TSGL-EEGNet is based on EEGNet which you can find it here.

Dataset

• BCI Competition IV 2a

You need to do some pre-processing works before using it. Maybe you need MATLAB and EEGlab tools to read these datas and transform them to *.mat file format.

Features

Framework for EEG

• Cross Validation Framework
• Grid Search Framework
• Cropped Training with CV and GS
• Visualization Framework

Implemented Models

• EEGNet [1]
• TSGL-EEGNet
• DeepConvNet [2]
• ShallowConvNet [2]
• Multi-branch 3D CNN [3] Not supported yet.
• FBCSP

Upgrade Log

• v1.1
  • use a new code construction to do cropped training
  • use generator to train on a large dataset
  • rename core.train to core.training, to be different from .train
  • add maxmin normalization
  • separate code test and model test
  • fix some bugs
• known issues
  • cv testing and model ensemble (stacking) are not adapted
• next
  • use generator to load a large dataset

Usage

Requirement

• Python >= 3.6 && <= 3.8
• Python >= 3.7 (when your GPU is newer than Nvidia 30xx)
• tensorflow-gpu >= 2.0.0 && <= 2.3.0
• tensorflow >= 2.4.1 (pip) (when your GPU is newer than Nvidia 30xx)
• scikit-learn >= 0.21.3
• scipy >= 1.3.1
• numpy >= 1.17.3
• tensorflow-addons (pip)
• hdf5 >= 1.10.4
• h5py >= 2.9.0
• matplotlib >= 3.1.1 && <=3.3.4 (optional)
• pydot >= 1.4.1 (optional)
• graphviz >= 2.38 (optional)
• mne >= 0.20.7 (pip) (optional)
• braindecode == 0.2.0 (pip) (fbcsp) (optional)

It is recommended to use conda environment. tensorflow-addons Requirement Optional packages are for visualization

Quick Start

# training models
python train.py

# testing single model
python model_test.py

# testing CV models (not adapted)
# python model_cv.py

# ensemble models (not adapted)
# python model_ensemble.py

# stacking models (not adapted)
# python model_stacking.py

# visualization
python vis.py

Code Structure

TODO

Paper Citation

If you use the EEGNet model in your research and found it helpful, please cite the following paper:

@article{Lawhern2018,
    author={Vernon J Lawhern and Amelia J Solon and Nicholas R Waytowich and Stephen M Gordon and Chou P Hung and Brent J Lance},
    title={EEGNet: a Compact Convolutional Neural Network for EEG-based Brain–computer Interfaces},
    journal={Journal of Neural Engineering},
    volume={15},
    number={5},
    pages={056013},
    url={http://stacks.iop.org/1741-2552/15/i=5/a=056013},
    year={2018}
}

If you use the TSGL-EEGNet model in your research and find this Repository is helpful, please cite the following paper:

@article{Deng2021,
    author={Deng, Xin and Zhang, Boxian and Yu, Nian and Liu, Ke and Sun, Kaiwei},
    journal={IEEE Access},
    title={Advanced TSGL-EEGNet for Motor Imagery EEG-Based Brain-Computer Interfaces},
    year={2021},
    volume={9},
    number={},
    pages={25118-25130},
    doi={10.1109/ACCESS.2021.3056088}
}

License

MIT License