Feature Compression Using PCA on Motor Imagery Classifications
Proceedings of 3rd International Conference on Internet of Things and Connected Technologies (ICIoTCT), 2018 held at Malaviya National Institute of Technology, Jaipur (India) on March 26-27, 2018
7 Pages Posted: 3 May 2018
Date Written: April 20, 2018
The electrical signals were first discovered by English scientist Richard Caton in 1875. Since then technology has evolved in all its spheres to alleviate the developments in this field. Nowadays with the help of well-equipped Brain-Computer Interface (BCI) a channel can be established between the human brain and immobile body parts. At the time of around 1920’s the study of electrical activity of the brain begun including observing the patterns in these signals in different frequency ranges. Furthermore, in this context electrophysiological monitoring of brain’s electrical signals splattered this domain. Electroencephalography (EEG) is a technique in which electrodes are placed along the scalp that measures voltage fluctuations that results because of the ionic current within the brain neurons. In this paper, classification of EEG signals is done which can be further used to equip a BCI system. One Such domain is Motor Imagery, where these signals are used to map the imaginative movements to the actual ones by translating the neural commands into control signals by classifying different EEG patterns observed. Analysis of “Two class motor Imagery” data is done which is procured via BNCI Horizon. Later, processing of data for feature acquisition is followed by classification. The data comprises of two class - right hand and limb for which data is classified using various machine learning algorithms.
Keywords: CV: Cross fold Validation; PS: Percentage Split; EEG: Electroencephalography; BCI: Brain Computer Interface
Suggested Citation: Suggested Citation