English Character recognition using EEG-based Visual stimulations: A Machine Learning Analysis

The development of Brain Computer Interfaces (BCI) using electroencephalography (EEG) signals in e-learning systems for disabled people has received much interest from researchers in the past decade. EEG signals are sensitive and can be altered due to different visual stimuli, such as color and shape of objects. The goal of this study is to develop a novel methodology to decode the English alphabets directly from EEG signals. The methodology consisted of EEG signal acquisition in a custom experimental condition, feature extraction and classification. EEG signals were collected from three subjects, who were presented English Alphabets in a white text in a dark environment. Based on literature, only the alpha wave was considered from the EEG signal. Some statistical features have been extracted using time-domain analysis, and three different machine learning approaches, i.e., support vector machines (SVM), K-nearest neighbors (K-NN), and Naive Bayes classifier have been used to classify the EEG signals corresponding to the different alphabets. Among three classification algorithms, K-NN has outperformed the other classification models with an accuracy of 81.6%. The developed method has several implications, including rehabilitation and brain-controlled automation.