python-gcpds.EEG_Tensorflow_models

Deep learning models applied to EEG signals on tensorflow 2.x

Datasets

Name	Subjects	fs[Hz]	Classes
BCI2a	9	250	Left hand / Right hand / Feet / Toungue
GIGA17	52	500	Left hand / Right hand

State-of-the-art methods

• DeepConvNet: Deep Learning With Convolutional Neural Networks for EEG Decoding and Visualization.
• ShallowConvNet: Deep Learning With Convolutional Neural Networks for EEG Decoding and Visualization.
• EEGNet: A Compact Convolutional Neural Network for EEG-based Brain-Computer Interfaces.
• DMTL_BCI: EEG-Based Motor Imagery Classification with Deep Multi-Task Learning.
• PST-Attention: Parallel Spatial–Temporal Self-Attention CNN-Based Motor Imagery Classification for BCI.
• TCNet-Fusion: Electroencephalography-based motor imagery classification using temporal convolutional network fusion.

Proposed approaches

• MTVAE: Multi-task variational autoencoder.
• ShallowConvNet_1Conv2d: Variation of ShallowConvNet with only a convolutional 2D layer.
• ShallowConvNet_1Conv2d_rff: : Variation of ShallowConvNet_1Conv2D using Random Fourier Features layer.
• Deep&Wide Learning using Gradient Class Activation Maps for Enhanced Physiological Interpretability of Motor Imagery Skills
• CNN_DW_ITL
• Preprocessing_MI

Install

1. Clone this repo.

git clone https://github.com/UN-GCPDS/python-gcpds.EEG_Tensorflow_models

2. Install repo.

On personal laptop

pip install -e git+https://github.com/UN-GCPDS/python-gcpds.EEG_Tensorflow_models.git#egg=EEG_Tensorflow_models

On Google Colab

pip install -U git+https://github.com/UN-GCPDS/python-gcpds.EEG_Tensorflow_models.git

Results

• BCI2a:

  

• GIGA17: