A time-frequency map generation network embedded with spectral self-attention for motor imagery classification

Xu Niu, Na Lu, Ruofan Yan, Huan Luo

Abstract: Highlights•TFGNet is developed as the first end-to-end network capable of leveraging spectral information from raw EEG signals to generate time–frequency maps.•A spectral attention module is first developed for end-to-end network of EEG signal analysis.•The time–frequency maps generated by TFGNet exhibit clear and complete ERDS patterns, as shown by the visualization results.•The weights assigned to samples by the attention module reveal individual differences in ERDS-based spectral patterns.