EEG-based Attention Grading and Channel Redundancy Testing

With the rapid development of brain-computer interface technology, the study of attention grading based on EEG signals has become a hot research issue within the field. In different studies, the EEG signal acquisition paradigm has no uniform objective evaluation criteria and is limited with EEG acquisition equipment, which makes it less feasible to be applied to real life. To address the above issues, we used the Schulte grid paradigm to achieve the acquisition and annotation of EEG data of different attention types; then we conducted visualisation experiments on the attention task, and selected the channels that contributed more to the recognition of attention features by observing the model's degree of attention to different channels. Finally, we conducted a reduction experiment on the EEG signal channels. The results show that: 1. most of the channels contributing most to the attention grading features are located in the right frontal region; 2. reducing the EEG signal channels to 5 channels, the recognition accuracy reaches up to 95.85% when comparing the existing attention grading algorithms using CNN-ECA-LSTM. This provides a reference for improving the portability and robustness of EEG measurement devices and increasing the possibility of practical applications.