GENETIC ALGORITHM (GA) BASED FEATURE SELECTION FOR ELECTROCORTICOGRAPHY (ECOG) BRAIN-COMPUTER INTERFACE (BCI)

  • Aswinseshadri.K, Dr.V.Thulasi Bai
Keywords: Brain Computer Interfaces (BCIs), Electroencephalography (EEG), Electrocorticography (ECoG), Wavelet Packet Transform (WPT), Genetic Algorithm (GA).

Abstract

Brain Computer Interfaces (BCIs) have given a way for the brain to be able to interface with the computer directly. There are different signals of the brain that are used for controlling a device that keeps varying in terms of noise, spatial resolution, temporal resolution, stability, reliability, and recording. Electrocorticography (ECoG) based electrodes have a reliable signal obtained from the human brain surface and such signals were used for decoding various movements, speech, and vision. The aim of feature selection is to select a new feature subset from an original set of features that are based on redundancy and relevance. For the purpose of this work, the technique of Wavelet Packet Transform (WPT) has been employed for the removal of artifacts from the signals. The Genetic Algorithm had been employed in the feature selection process for extracting certain relevant classification features. The algorithm’s superiority was confirmed by the experimental results of the BCI Competition III dataset I. Theproposed feature classification approach was evaluated forSupport Vector Machine (SVM) Logistic Regression (LR) andRandom Forest (RF) in this work.

Published
2021-06-25
How to Cite
Dr.V.Thulasi Bai, A. (2021). GENETIC ALGORITHM (GA) BASED FEATURE SELECTION FOR ELECTROCORTICOGRAPHY (ECOG) BRAIN-COMPUTER INTERFACE (BCI). Design Engineering, 133-143. Retrieved from http://thedesignengineering.com/index.php/DE/article/view/2264
Section
Articles