Disciplines
Computational Engineering | Electrical and Electronics
Abstract (300 words maximum)
This paper investigates the use of deep learning as a means for quantification and source localization of prioritizing electroencephalogram (EEG) waves for the purpose of detecting different eye states of human subjects. The Convolutional Deep Learning tool is trained to recognize EEG reading corresponding to a set of different eye movements as generated by watching different action scenes. The results also predict whether the subjects' eyes are open or closed. Source localization is performed next on the EEG data to focus on the different EEG components which primarily contribute to the activity. This was done by using a convolutional neural network to determine the brain region where the stimulation occurred. Finally, based on the selected region of primary stimulus of the brain as received by the EEG device, the residual EEG node data are removed as redundant. This significantly saves the computation load of the machine learning as it performs on a reduced dataset, without compromising on its performance efficiency. Training a machine to detect the location of an EEG wave and the intensity of said wave would have many applications in the medical and industrial fields.
Academic department under which the project should be listed
SPCEET - Electrical and Computer Engineering
Primary Investigator (PI) Name
Sumit Chakravarty
Source Localization of Electroencephalogram (EEG) Waves with Convolutional Neural Network
This paper investigates the use of deep learning as a means for quantification and source localization of prioritizing electroencephalogram (EEG) waves for the purpose of detecting different eye states of human subjects. The Convolutional Deep Learning tool is trained to recognize EEG reading corresponding to a set of different eye movements as generated by watching different action scenes. The results also predict whether the subjects' eyes are open or closed. Source localization is performed next on the EEG data to focus on the different EEG components which primarily contribute to the activity. This was done by using a convolutional neural network to determine the brain region where the stimulation occurred. Finally, based on the selected region of primary stimulus of the brain as received by the EEG device, the residual EEG node data are removed as redundant. This significantly saves the computation load of the machine learning as it performs on a reduced dataset, without compromising on its performance efficiency. Training a machine to detect the location of an EEG wave and the intensity of said wave would have many applications in the medical and industrial fields.