/FRDW

This repository is the implementation of “Front-end Replication Dynamic Window (FRDW) for Online Motor Imagery Classification."

Primary LanguagePython

FRDW

This repository is the implementation of “Front-end Replication Dynamic Window (FRDW) for Online Motor Imagery Classification."

Usage Examples

Here is an example code snippet demonstrating how to use the project:

  1. within-subject_train: for finding the best parameters and training the models
python within-subject_train.py --lr 0.001 --gpu_id '2' --seed 2022 --epoch 180 --bs 64 --train_len 100 --model_num 2 --dataset '001-2014' --classes 2 --person 1 --augmentation overlap --overlap 25 --model_type Transformer --model_save_path /model/try/
  1. cross-subject_train: for finding the best parameters and training the models
python cross-subject_train.py --lr 0.001 --gpu_id '2' --seed 2022 --epoch 180 --bs 64 --train_len 100 --model_num 2 --dataset '004-2014' --classes 2  --person 1 --augmentation overlap --overlap 25 --model_type Transformer --model_save_path /model/try/ --if_EA True
  1. FW: fixed window (FW)
  2. FRDW: FRDW
python FW.py --seed 2022 --model_num 2 --train_length 100 --dataset '001-2014' --classes 4 --person 1 --model_type 'EEGNet' --model_save_path /model/EEGNet-001-2014-4/within_overlap/ --gpu_id '3'
python FRDW.py --seed 2022 --model_num 2 --train_length 100 --dataset '001-2014' --classes 4 --person 1 --model_type 'EEGNet' --model_save_path /model/EEGNet-001-2014-4/within_overlap/ --gpu_id '3'
  1. FW+EA: fixed window (FW) with EA
  2. FRDW+EA: FRDW with EA
python FW+EA.py --seed 2022 --model_num 2 --train_length 100 --dataset '001-2014' --classes 4 --person 1 --model_type 'EEGNet' --model_save_path /model/EEGNet-001-2014-4/cross_overlap/ --gpu_id '3' --modelEA_save_path /model/EEGNet-001-2014-4/cross_overlap_EA/
python FRDW+EA.py --seed 2022 --model_num 2 --train_length 100 --dataset '001-2014' --classes 4 --person 1 --model_type 'EEGNet' --model_save_path /model/EEGNet-001-2014-4/cross_overlap/ --gpu_id '3' --modelEA_save_path /model/EEGNet-001-2014-4/cross_overlap_EA/

Dataset

You can find the dataset at the following locations: http://www.bnci-horizon-2020.eu/database/data-sets or https://www.bbci.de/competition/iv/ .We have provided data processing code for both .gdf and .mat formats.

Installing Dependencies

pip install -r requirements.txt

if you're using conda, you can create and activate an environment:

conda env create -f environment.yml
conda activate my_environment