/EEGNeX

About Open source code of paper: Toward reliable signals decoding for electroencephalogram: A benchmark study to EEGNeX

Primary LanguagePythonMIT LicenseMIT

EEGNeX

GitHub license GitHub stars Twitter

About Open source code of paper:
-Toward reliable signals decoding for electroencephalogram: A benchmark study to EEGNeX.
-https://arxiv.org/abs/2207.12369


1

This notebook is released for easy implementation of running all benchmarkmodels designed for electroencephalography(EEG) classification tasks, including:

  • Single_LSTM
  • Single_GRU
  • OneD_CNN
  • OneD_CNN_Dilated
  • OneD_CNN_Causal
  • OneD_CNN_CausalDilated
  • TwoD_CNN
  • TwoD_CNN_Dilated
  • TwoD_CNN_Separable
  • TwoD_CNN_Depthwise
  • CNN_LSTM
  • CNN_GRU
  • Single_ConvLSTM2D
  • EEGNet_4_2
  • EEGNet_8_2
  • EEGNeX_8_32

For running the code, please run notebook Run_model.ipynb

Additional python packages required:

  • keras 2.8.0
  • tensorflow 2.8.0
  • torch 1.10.2

The result folder contains validation results of running benchmarkmodels on four EEG datasets from paper.

4
EEGNeX architecture

5


More models are planned to be added:

  • DeepConvNet
  • ShallowConvNet
  • SNN(Spike neural network)_based models

We also welcome you to contribute any model resources/papers in the discussion for our future plan :)