Students' Attention Classification During Class Lecture Using BCI and Machine Learning: A Pathway Towards Neurofeedback-Based Learning

Students' effective learning during a class depends on their level of attention to the lecture. Students who tend to be more attentive during class lectures are more likely to perform better. However, day-by-day more and more students are failing to be attentive in class and their attention span has become lower than ever before. By using students' attention in class an automatic Brain Computer Interface (BCI) based self-feedback system might become revolutionary in education and learning. But separating inattentive students from attentive students using BCI is a problem in itself. In this study, we designed an experiment to classify between attentive students and inattentive students using BCI with machine learning and utilizing brain-to-brain synchronization theory. We designed an experiment where we collected EEG data from 9 students and 1 teacher during a class lecture. During the lecture, the students were asked to fill out three types of questionnaires which separated them into three attention groups. After pre-processing the data, 473 features were extracted and 50 were selected using recursive feature elimination (SVM-RFE). Finally, the data were classified using SVM and 80.64% accuracy was achieved between the attentive and inattentive groups. The outcome is a huge leap forward to build the world's first neurofeedback-based learning system.