EEG Conformer: Convolutional Transformer for EEG Decoding and Visualization

By Yonghao Song, Qingqing Zheng, Bingchuan Liu, Xiaorong Gao. IEEE transactions on neural systems and rehabilitation engineering, 2022

981 citations recorded by the source corpus.

Due to the limited perceptual field, convolutional neural networks (CNN) only extract local temporal features and may fail to capture long-term dependencies for EEG decoding. In this paper, we propose a compact Convolutional Transformer, named EEG Conformer, to encapsulate local and global features in a unified EEG classification framework. Specifically, the convolution module learns the low-level local features throughout the one-dimensional temporal and spatial convolution layers. The self-attention module is straightforwardly connected to extract the global correlation within the local temporal features. Subsequently, the simple classifier module based on fully-connected layers is followed to predict the categories for EEG signals. To enhance interpretability, we also devise a visualization strategy to project the class activation mapping onto the brain topography. Finally, we have conducted extensive experiments to evaluate our method on three public datasets in EEG-based motor imagery and emotion recognition paradigms. The experimental results show that our method achieves state-of-the-art performance and has great potential to be a new baseline for general EEG decoding. The code has been released in https://github.com/eeyhsong/EEG-Conformer.

Subfields: Cognitive Neuroscience

View Paper: https://www.semanticscholar.org/paper/db10636e62862c9a4bd3e75012ae0273492ec125

Citations in the corpus