Structure invariance-driven collaborative contrastive network for EEG decoding

Shuang Liang, Zengguang Li, Wenlong Hang, Yuanpeng Zhang, Baiying Lei, Jing Qin, Yu Zhang, Kup-Sze Choi

Abstract: Highlights•Novel structure invariance-driven collaborative contrastive network (SICCNet) for EEG decoding.•Novel EEG DA method linearly interpolates the segments of the same class data to alleviate the undesirable noise.•SICCNet learns the invariant data-structure information across augmented views to enhance the generalizability of the model.