Επεξεργασία και ανάλυση ηλεκτροεγκεφαλογραφικών σημάτων για την αναγνώριση συγκέντρωσης του υποκειμένου

Abstract

In the last years, the use of Electroencephalogram (EEG) signals in the field of Machine Learning has obtained a lot of interest, since it would be very useful to deeply understand these signals. In this context, the aim of the present thesis is to create an algorithm, which will use humans’ EEG signals in order to recognize whether or not these humans are concentrated during the measurement (Attentiveness Recognition System). To develop this system, an open access dataset of EEG signals was used. These signals were processed digitally so they are suitable to be fed in the Pattern Recognition System that was built for the purpose of this thesis. This System analyzes the EEG signals, employing Feature Extraction and Feature Selection techniques, in order to classify them depending on the concentration of the volunteers during the EEG measurement. Thus, the data were classified into two classes; the signals which correspond to volunteers who were concentrated belong to the first class, and the signals which correspond to volunteers who were in relaxed-not concentrated state, belong to the second class. The results of the System show that it is possible to successfully discriminate and classify EEG signals according to the concentration of the subject.