Eye State Classification Through Analysis of EEG Data Using Deep Learning

The purpose of this study is to create a network that can detect the state at which the eyes are in at a specific time step through analysis of a dataset recorded using a 14-channel Emotiv EEG Neuroheadset. This study can be useful and serve as a supporting input in the development of other researches and systems that considers eye state and movement as an important factor and input, such as driving state detection projects specifically the classification of drowsiness levels. In this paper, deep learning was applied in creating the network, trained with a total of 10,424 data points, validated to classify only two states: eyes open and eyes closed. The network was trained and completed using MATLAB and Microsoft Excel. Accuracy of the classification action between the testing data and the completed output network in this study achieved 89.23% across all 4,468 data points.