Eye State Detection from Electro-Encephalography Signals using Machine learning Techniques

Detection of the Eye State can be helpful in several ways since it can be used as an indication of mental disorders, energetics in aging, detection of alcoholism, and detection of driver's drowsiness. It can also detect seizures in people suffering from Epilepsy as during the seizure attacks, there is a rapid eye blinking rate. Thus, the detection of Eye State is an important achievement. This paper aims to analyze Electroencephalograph (EEG) signals to detect the state of the human eye, ie eye closed or eye open, with the help of machine learning techniques with an optimized number of EEG Channels. Here, different machine learning techniques such as Support Vector Machine (SVM), K- Nearest Neighbor (KNN), and Random Forest, along with open source datasets are used for the identification. Eye state detection, with 14 channels, with 80-20 train-test ratio, is achieved with 100%, 95.73%, 100% training accuracy, and 98.18%, 97.54% 92.34% test accuracy by SVM, KNN, and Random Forest algorithms, respectively. Whereas for the same techniques and same train-test ratio, 10 channels and 7 channels both training and testing accuracies are more than 88%. Eye open- eye closed state detection using three machine learning techniques is carried out with 14 channels, 10 channels, and 7 channels.