AUTHOR=Li Penghai , Su Jianxian , Belkacem Abdelkader Nasreddine , Cheng Longlong , Chen Chao 

TITLE=Multi-person feature fusion transfer learning-based convolutional neural network for SSVEP-based collaborative BCI

JOURNAL=Frontiers in Neuroscience

VOLUME=Volume 16 - 2022

YEAR=2022

URL=https://www.frontiersin.org/journals/neuroscience/articles/10.3389/fnins.2022.971039

DOI=10.3389/fnins.2022.971039

ISSN=1662-453X

ABSTRACT=Objective: To solve typical problems in the existing single-person brain–computer interface systems such as low signal-to-noise ratio, distinct individual differences, and unstable execution effect, a centralized and collaborative steady-state visually evoked potential brain–computer interface system (SSVEP-cBCI), which can fuse multi-person electroencephalography (EEG) features, was constructed. Furthermore,  three methods for multi-person feature fusion were developed and applied to EEG classification. A comparative analysis of their classification accuracy was performed with convolutional neural network (CNN)-based on transfer learning (TL).
Approach: An EEG-based SSVEP-cBCI system was developed, which merges different users’ EEG features stimulated by the instructions for the same task. Three multi-person features fusion methods, namely, parallel connecting, serial connecting, and multi-person averaging, were applied. The fused features were then fed as input to CNN for classification. The TL method was applied to a THU benchmark standard dataset and a self-collected dataset, to effectively reduce the size of self-collected dataset required for CNN training and increase the classification accuracy. Ten subjects were recruited for data collection, and the collected data were used to gauge the performance of the three fusion algorithms.
Main results: The results predicted by TL-CNN in single-person mode and the three feature fusion methods of multi-person mode were compared. The experimental results show that each multi-person mode is superior to single-person mode. Within the three-second time window, the classification accuracy of the single-person CNN is only 90.6%, while the classification accuracy of the feature fusion method of two-person parallel connecting can reach 96.6%, achieving better classification results.
Significance: The results show that the three multi-person feature fusion methods and deep learning classification algorithm based on multi-person feature fusion TL-CNN can effectively improve the SSVEP-cBCI classification performance. The feature fusion method of two-person parallel feature connecting achieves better classification results. Different feature fusion methods can be selected in different application scenarios to further optimize cBCI.