A learnable continuous wavelet-based multi-branch attentive convolutional neural network for spatio-spectral-temporal EEG signal decoding

Junmo Kim, Keun-Soo Heo, Dong-Hee Shin, Hyeonyeong Nam, Dong-Ok Won, Ji-Hoon Jeong, Tae-Eui Kam

Abstract: Highlights•Motor Imagery-EEG decoding plays a critical role in neurorehabilitation and BCI.•We introduce a novel method for spatio-spectral–temporal motor imagery-EEG decoding.•Our method transforms EEG into discriminative SST representations automatically.•Our multi-branch attentive CNN learns local/global SST features simultaneously.•We demonstrate its effectiveness on public datasets by comparing with SOTA methods.