Development of EEG classifier based on EEG data grouped by SOM

Abstract

In recent years, researches on BMI have attracted much attention. In previous studies, indi- vidual EEG classifiers were used generally since the brain waves are different for each individual. However, creating an EEG classifier for each subject causes an increase in data collection time and calculation cost. In our previous study, we tried to construct a common EEG classifier based on EEG data measured from multiple subjects, but the classification rate was not improved. In this research, it is assumed that there exist a suitable subject group and unsuitable subjects for the common EEG classifier, and EEG data are divided into such groups using SOM. A common classifier for the suitable subject group is constructed and compared with previous classifier.