EEG Signal Classification with Machine Learning model using PCA feature selection with Modified Hilbert transformation for Brain-Computer Interface Application

Abstract

Brain Computer Interfaces (BCI’s) computes the communication and control channels that rely on the output channel normal brain for the peripheral nerves. The present BCI advancement incorporates the technologies for augmentative communication between the users for the conventional technique for the effective control of muscles. The application of the Electroencephalography (EEG) activity estimates the brain activity for effective communication technology. This paper presented a Principal Component Analysis (PCA) integrated with the Modified Hilbert Transformation defined as the PCAHT. The developed model uses the PCA model for the estimation of the feature selection in the EEG signal for processing. The EEG signals inputs are pre-processed and evaluated for the features within the signal for processing. The proposed PCAHT model incorporates the modified Hilbert Transformation model for signal conversion and is applied over the machine learning model. The deployed PCAHT model perform the classification of the EEG signal those are normal and abnormal activity in the brain signals. The simulation analysis expressed that proposed PCAHT model achieves a higher accuracy value of 98%, recall value is measured as 98% and precision value is measured as the 98%. The comparative analysis observed that the proposed PCAHT model achieves the ~4% improved performance than the conventional classifier models.