EMD-Based Feature to Detect Situational Interest in Classroom Settings Using EEG

Abstract

Student's performance assessment tools had gone through a huge improvement throughout time and reached high levels of reliability and accuracy. According to several studies, situational interest measure can be used to assess and improve student's academic achievement. So far, there is a lack in defining objective measures for situational interest rather than depending on the subjective measures. In this paper, we propose the use of EMD-based feature to detect situational interest in classroom settings using EEG. Several subjective measures: Personal Interest (PI) test, Situational Interest (SI) test and knowledge tests were used to evaluate subjectively student's interest. EEG data were then preprocessed and decomposed using Empirical Mode Decomposition (EMD) and the resulted Intrinsic Mode Functions (IMFs) was ranked using t-test to select the most significant ones. Eight features out of the 64 features extracted, were then fed into KNN and SVM classifiers achieving high accuracy of 90% and 100% respectively using features from temporal regions.