Motor imagery classification based on variable precision multigranulation rough set

Abstract

In this work classification based on Variable Precision Multigranulation Rough Set for motor imagery dataset is proposed. The accurate design of BCI (Brain Computer Interface) depends upon efficient classification of motor imagery movements of patients. In the first phase pre-processing is carried out with Chebyshev type2 filter in order to remove the noises that may exist in signal during acquisition. The daubechies wavelet is used to extract features from EEG Signal. Finally classification is done with Variable Precision Multigranulation Rough Set. An experimental result depicts higher accuracy according to variation of alpha and beta values in Variable Precision multigranulation rough set.