End-to-End Multi-Task Learning for Subject-Independent Motor Imagery EEG Classification

Python API and the novel algorithm for motor imagery EEG recognition named MIN2Net. The API benefits BCI researchers ranging from beginners to experts. We demonstrate the examples in using the API for loading benchmark datasets, preprocessing, training, and validation of SOTA models, including MIN2Net. In summary, the API allows the researchers to construct the pipeline for benchmarking the newly proposed models and very recently developed SOTA models.

Website: https://min2net.github.io
Documentation: https://min2net.github.io
Source code: https://github.com/IoBT-VISTEC/MIN2Net
Bug reports: https://github.com/IoBT-VISTEC/MIN2Net/issues

📌 PyTorch Version 🔁

The MIN2Net architecture implemented in PyTorch is available on:

👉 AlphaGrad Repository – mtl_bci/networks/MIN2Net.py

Getting started

Dependencies

Python==3.6.9
tensorflow-gpu==2.2.0
tensorflow-addons==0.9.1
scikit-learn>=0.24.1
wget>=3.2

Create conda environment with dependencies

wget https://raw.githubusercontent.com/IoBT-VISTEC/MIN2Net/main/environment.yml
conda env create -f environment.yml
conda activate min2net

Installation:

Using pip

pip install min2net

Using the released python wheel

wget https://github.com/IoBT-VISTEC/MIN2Net/releases/download/v1.0.1/min2net-1.0.1-py3-none-any.whl
pip install min2net-1.0.1-py3-none-any.whl

Tutorial

<img src="https://min2net.github.io/assets/images/colab_favicon.ico" width="50" height="50"> Open in Colab

Citation

To cited our paper

P. Autthasan et al., "MIN2Net: End-to-End Multi-Task Learning for Subject-Independent Motor Imagery EEG Classification," in IEEE Transactions on Biomedical Engineering, doi: 10.1109/TBME.2021.3137184.

@ARTICLE{9658165,
  author={Autthasan, Phairot and Chaisaen, Rattanaphon and Sudhawiyangkul, Thapanun and Rangpong, Phurin and Kiatthaveephong, Suktipol and Dilokthanakul, Nat and Bhakdisongkhram, Gun and Phan, Huy and Guan, Cuntai and Wilaiprasitporn, Theerawit},
  journal={IEEE Transactions on Biomedical Engineering}, 
  title={MIN2Net: End-to-End Multi-Task Learning for Subject-Independent Motor Imagery EEG Classification}, 
  year={2022},
  volume={69},
  number={6},
  pages={2105-2118},
  doi={10.1109/TBME.2021.3137184}}

MIN2Net

Install / Use

README