Motor Imagery Brain Activity Recognition through Data Augmentation using DC-GANs and Mu-Sigma

Abstract

The brain-computer interface is a technology that allows a machine to connect with the human brain and work based on the commands released by thoughts and activities of the brain. Electrodes are placed on the scalp and the changes in electric waves released by the brain are recorded as Electroencephalography (EEG) signals. In this work, we propose the use of generative adversarial networks and musigma methods to augment the EEG signals. Some of the existing deep learning methods such as convolutional neural network and recurrent neural network for classification of the EEG signals are implemented and their classification performance is examined with and without data augmentation. It is shown that the use of data augmentation can improve the performance of the EEG signal classification with deep learning models to a considerable extend.