This is the PyTorch implementation of the temporal dependency learning CNN for MI-EEG decoding.
The temporal dependency learning CNN is designed with the aim of learning temporal dependencies between discriminative features in different time periods during MI tasks. It is composed of the following five stages:
This repository is designed as a toolbox that provides all necessary tools for training and testing the proposed network. All the data functionalities are defined in the data directory. After preprocessing data, the cv.py and train_test.py are the entry points to train and test the proposed network in the session-dependent setting and session-independent setting (defined in the paper), respectively.
The classification results for our proposed network and other competing architectures are as follows:
If you find this code useful, please cite us in your paper.
@article{ma2023temporal,\ title={A Temporal Dependency Learning CNN With Attention Mechanism for MI-EEG Decoding},\ author={Ma, Xinzhi and Chen, Weihai and Pei, Zhongcai and Liu, Jingmeng and Huang, Bin and Chen, Jianer},\ journal={IEEE Transactions on Neural Systems and Rehabilitation Engineering},\ year={2023},\ volume={31},\ pages={3188-3200},\ doi={10.1109/TNSRE.2023.3299355}\ }