Multi-brain Collaborative Target Detection Based on RAP

Brain-computer interfaces establish direct connections between the human brain and external devices, enabling information exchange. In the auditory pathway, a BCI system based on Rapid Auditory Presentation (RAP) can detect Event-Related Potentials (ERP) in electroencephalogram (EEG) signals and use machine learning or deep learning methods to achieve auditory target detection in Oddball sequences. However, due to the low amplitude of auditory evoked potential and the low signal-to-noise ratio of signals, the accuracy of target recognition is often low and this makes it difficult to meet application demands. Therefore, this paper proposes a multi-brain collaborative target detection method based on RAP. Firstly, this article designs two brain-computer interface experimental paradigms based on rapid auditory presentation for EEG signal acquisition. Secondly, a feature extraction method combining downsampling and mean filtering is used to extract time-domain features from segmented data. Then, three different classifiers are used to train and predict the experimental data, and multi-brain information fusion is performed for the predicted results as the final result. Finally, the real-time performance and target detection effect of the proposed classification discriminative model are verified from the target stimulus recall rate and information transmission rate. The experimental results show that the recognition accuracy and information transmission rate of the multi-brain information fusion strategy adopted in this paper are improved by 23.42% and 7.10 Bit/min, respectively. This indicates that the proposed approach is effective in improving the real-time performance and target detection accuracy of the RAP-BCI system.