run.md

Run

Environment

• Python 3.9

conda create -n atdgnn python=3.9

conda activate atdgnn

Install Requirements

pip3 install -r requirements.txt

Run

python main.py

You can set the parameters in main.py to run the code. You can also use flags to set the parameters. The parameters used in our paper are listed in the params file.

python main.py --label 'V'

DEAP

If you want to run the code with the DEAP dataset, you must set these parameters as follows:

python main.py --dataset 'DEAP' --sampling-rate=128 --target-rate=128 --trial-duration=63 --input-shape '1, 32, 512'

For other parameters, you can refer to the params file.

MEEG

If you want to run the code with the MEEG dataset, you must set these parameters as follows:

python main.py --dataset 'MEEG' --sampling-rate=1000 --target-rate=200 --trial-duration=59 --input-shape '1, 32, 800'

For other parameters, you can refer to the params file.

Example

• Clone our code.

  git clone https://github.com/xmh1011/AT-DGNN.git
  cd AT-DGNN

• Unzip the data file.

  cd example
  tar -xvzf s01.tar.gz
  tar -xvzf sample_1.tar.gz

example/sample_1.dat is subject 1 of MEEG dataset. example/s01.dat is subject 1 of DEAP dataset.

After unzipping the sample data, you can run the code as follows.

MEEG

example/sample_1.dat is subject 1 of MEEG dataset.

You can run as follows:

python main.py --data-path './example' --dataset 'MEEG' --sampling-rate=1000 --target-rate=200 --trial-duration=59 --input-shape '1,32,800' --subjects=1 --model 'AT-DGNN'

DEAP

example/s01.dat is subject 1 of DEAP dataset.

You can run as follows:

python main.py --data-path './example' --dataset 'DEAP' --sampling-rate=128 --target-rate=128 --trial-duration=63 --input-shape '1,32,512' --subjects=1 --model 'AT-DGNN'

Reproduce

If you want to reproduce the results in our paper, you can download the dataset. After downloading the dataset, you can run the code as mentioned above.