Abstract
In P300 speller, which is one of Brain Computer Interface (BCI), online experiment is carried out to show a statistical difference between methods. However, it is difficult to show a statistical difference unless many subjects are involved in an experiment. Thus we have carried out offline experiment so far. In the offline experiment, we randomly pick up some data from those measured by online experiment(s) repeatedly as sampling with replacement and compare the methods many times. It is not difficult to show a statistical difference between methods because we can repeat the trials as many times as desired by using same data without considering how much difference there is. This paper studies the offline experiment based on the reliability of data.
