Impact of Feature Selection on EEG Based Motor Imagery

Abstract

An EEG based motor imagery translates the motor intention of any subject into control signal by classifying EEG data of different imagination tasks such as hand and feet movements. As indicated by study, it is found that there are almost around 1 in 50 individuals living with loss of motion roughly 5.4 million individuals. For this sort of inability, EEG based BCI for motor imagery of right hand and feet movement imagination is acquired and classified. Short time Fourier transform and wavelet features are extracted and classified with and without feature selection. Ranking method is used for feature selection. Both classification outcomes are comparatively analyzed and observed that there is an increment in classification accuracy when features are classified after feature selection.