XAI606

This repo contains the non-official implementation of EEGNet pytorch version.

1. Installation

Environment

Python == 3.7.10
Pytorch == 1.9.0
CUDA 11.0

Dependencies

Create conda environment

conda == 4.10.1

(Option 1) Using yaml file

conda env create --file xai606.yaml

(Option 2) Install packages manually

conda install pytorch=1.9.0 cudatoolkit=11.1 -c pytorch -c nvidia
conda install numpy pandas matplotlib pyyaml ipywidgets
pip install torchinfo

2. Directory structure

.
├── README.md
├── base
├── base_trainer.py
│   └── layers.py
├── configs
│   └── EEGNet_config.yaml
├── data_loader
│   ├── __pycache__
│   ├── data_generator.py
│   └── dataset
├── figures
│   ├── dataset.png
│   └── directory_structure.png
├── history.ipynb
├── main.py
├── models
│   ├── EEGNet_model.py
│   └── model_builder.py
├── runs
│   ├── evaluation.sh
│   ├── prediction.sh
│   └── train.sh
├── trainers
│   ├── EEGNet_trainer.py
│   └── trainer_maker.py
├── utils
│   ├── __pycache__
│   ├── calculator.py
│   ├── get_args.py
│   ├── logger.py
│   └── utils.py
└── xai606.yaml

3. Dataset

.
├── test
│   ├── S01_X.npy
│   ├── S02_X.npy
│   ├── S03_X.npy
│   ├── S04_X.npy
│   ├── S05_X.npy
│   ├── S06_X.npy
│   ├── S07_X.npy
│   ├── S08_X.npy
│   ├── S09_X.npy
├── train
│   ├── S01_X.npy
│   ├── S01_y.npy
│   ├── S02_X.npy
│   ├── S02_y.npy
│   ├── S03_X.npy
│   ├── S03_y.npy
│   ├── S04_X.npy
│   ├── S04_y.npy
│   ├── S05_X.npy
│   ├── S05_y.npy
│   ├── S06_X.npy
│   ├── S06_y.npy
│   ├── S07_X.npy
│   ├── S07_y.npy
│   ├── S08_X.npy
│   ├── S08_y.npy
│   ├── S09_X.npy
│   └── S09_y.npy
└── val
    ├── S01_X.npy
    ├── S01_y.npy
    ├── S02_X.npy
    ├── S02_y.npy
    ├── S03_X.npy
    ├── S03_y.npy
    ├── S04_X.npy
    ├── S04_y.npy
    ├── S05_X.npy
    ├── S05_y.npy
    ├── S06_X.npy
    ├── S06_y.npy
    ├── S07_X.npy
    ├── S07_y.npy
    ├── S08_X.npy
    ├── S08_y.npy
    ├── S09_X.npy
    └── S09_y.npy

BCI Competition IV-2a dataset

9 subjects
Classes: left hand, right hand, feet, tongue (4 classes)
Session-to-session set up (=subject dependent)
Training set: 216 trials per subject
Validation set: 72 trials per subject
Test set: 288 trials per subject

Preprocessing

Sampling rate: 250Hz
Time segment: [0.5, 2.5]s post-cue
Band-pass filtering: 0-38Hz
Normalization: exponential moving average

4. Experiments

Models	S01	S02	S03	S04	S05	S06	S07	S08	S09	Mean
EEGNet	76.74	54.51	79.17	54.51	63.19	57.64	83.68	75.00	68.40	68.09

5. Get started

Train

sh runs/train.sh

Prediction

sh runs/prediction.sh

Visualization

Please note that history.ipynb file

6. Submission

./result/{save_dir}/{sub_dir}/prediction 폴더를 이메일로 제출해주세요.
혹은 S01~S09 각각의 prediction이 담겨있는 폴더를 제출해주세요 (확장자 무관).
donghee-ko@korea.ac.kr

High-East/XAI606-EEGNet