Cross Evaluation for Characteristics of Motor Imagery Using Neuro-feedback Based EEG-Brain Computer Interface

摘要

In this paper, we evaluated BCI algorithm using CSP for finding out about realistic possibility of BCI based on neurofeedback. BCI algorithm that was comprised of CSP and 3 kinds of classifier such as least square linear classifier, SVM and linear discriminant analysis was evaluated in 10 people. According to the result of the experiment, the effect of neurofeedback is evaluated. And in case of neurofeedback, some subject is exceptional but general trend shows the performance improvement by neurofeedback. This study gives the need for adaptive evaluation of motor imagery using neurofeedback based EEG-BCI that can be related to generic and robust system for automated subject-specific classification of EEG based BCI system, development of EEG based BCI literacy performance evaluation system.