A Study on Classification of EEG Data using Individual Classifiers of Other Subjects

Abstract

In recent years, studies on BMI and BCI, which use EEG as an interface for operating external devices, has been actively conducted. In many of those studies, an electroencephalogram classifier individualized to one subject has been created from the electroencephalogram data of the subject. This is because EEG data of one subject is said to be quite different from other subjects, and an individual classifier has advantage for increasing the classification rate. In general, a large amount of data is required for constructing the individual classifier. As a result, the burden of subjects increases. In this study, we input a small amount of EEG data of different subjects into the pre-built individual classifiers for multiple other subjects and compare the classification rates. We show that, without building a classifier individualized to one subject, there is a possibility to be able to use other subject’s individual classifier that has already been built.