A Subject-Transfer Study on Detecting c-VEP

Among all kinds of brain-computer interfaces(BCI), the BCI based on code visual evoked potential(c-VEP) distinguishes itself in stability and reliability. Generally, it is necessary to acquire training data for each subject before using c-VEP BCI. In order to facilitate the preparatory procedure, we have conducted a systematic research on subject-transfer problem about the detection of c-VEP. In this work, we proposed a similarity metric by which we can select the most suitable subjects for a new subject from our database, and developed a method to transfer a classifier from the selected subjects to detect the c-VEP of the new subject. We have verified our scheme using the data of 8 subjects. The results show that the classifiers that are transferred from other subjects work as well as those obtained by training on their own data.