“…The mainstream seizure detection methods are based on deep neural networks due to their high accuracy and end-to-end computation. Commonly used network architectures include the convolutional neural network (Hu et al, 2018 ; Wei et al, 2019 ; Abiyev et al, 2020 ; O'Shea et al, 2020 ; Ke et al, 2021 ; Shen et al, 2023 ), recurrent neural network (RNN; Abdelhameed et al, 2018 ; Hu et al, 2020 ), graph neural network (Wang et al, 2020 ; He et al, 2022 ; Tang et al, 2022 ), Transformer (Ke et al, 2022 ; Sun et al, 2022 ), and their combination (Abdelhameed et al, 2018 ; Jia et al, 2020 ; Ke et al, 2022 ). However, these kinds of networks treat EEG signals as image-liked inputs, which may not better utilize biological information.…”