Improvement in Classification of Tactile Event-Related Potentials using Random-Interval Tasks∗.

Abstract

The classification accuracy of an event-related potential (ERP) induced by a tactile stimulus is investigated to create a brain-computer interface (BCI). Mechanical tactile stimuli were applied to the left and right index fingers using two piezo actuator probes that were used as a Braille display for the visually impaired. In an experiment, two-class classification was investigated using three kinds of tactile stimulus pattern. The subjects were instructed to pay attention to unusual target stimuli while avoiding other frequent nontarget stimuli. The extracted features were classified using stepwise linear discriminant analysis. As a result, high accuracy was obtained by the task of random intervals compared with the task of constant intervals. It was suggested that the accuracy of the BCI using tactile stimuli is influenced by the concentration on the task.