This repository contains the code for RMCNN implemented with PyTorch.
More details in paper: A Riemannian Convolutional Nueral Network for EEG-Based Motor Imagery Decoding
Implementations of FBCSP-SVM, FBCNet, FBMSNet, Conformer, TSFCNet, Tensor-CSPNet, Graph-CSPNet and MAtt
All these benchmark methods are implemented in Pytorch.
FBCSP-SVM is provided at https://github.com/fbcsptoolbox/fbcsp_code
FBCNet is provided at https://github.com/ravikiran-mane/FBCNet
FBMSNet is provided at https://github.com/ravikiran-mane/FBCNet
Conformer is provided at https://github.com/eeyhsong/EEG-Conformer
TSFCNet is provided at https://github.com/hongyizhi/TSFCNet
Tensor-CSPNet and Graph-CSPNet is provided at https://github.com/GeometricBCI/Tensor-CSPNet-and-Graph-CSPNet
MAtt is provided at https://github.com/CECNL/MAtt
- model - This file contains the model used in this repository.
- utils - This file contains the functions used in this repository.
- FBCSP-SVM - This file contains the example code for classifying MI-EEG data using FBCSP-SVM.
- Tensor-CSPNet and Graph-CSPNet - This file contains the example code for classifying MI-EEG data using Tensor-CSPNet or Graph-CSPNet.
- main_FBCNet.py - An example code for classifying MI-EEG data using FBCNet.
- main_FBMSNet.py - An example code for classifying MI-EEG data using FBMSNet.
- main_Conformer.py - An example code for classifying MI-EEG data using Conformer.
- main_TSFCNet .py - An example code for classifying MI-EEG data using TSFCNet.
- main_MAtt.py - An example code for classifying MI-EEG data using MAtt.
- hold_out_benchmark.py - holdout code for FBCNet, FBMSNet, Conformer, TSFCNet and MAtt.
- The BCIC-IV-2a dataset can be downloaded in the following link: https://www.bbci.de/competition/iv/.
- The OpenBMI dataset can be downloaded in the following link: http://gigadb.org/dataset/100542.
- The High GAMMA dataset can be downloaded in the following link: https://gin.g-node.org/robintibor/high-gamma-dataset.