An Accurate Brainwave-Based Emotion Clustering for Learning Evaluation

Abstract

The purpose of this study is to help teachers understand their students’ learning situation. Especially in engineering education, Project-based Learning (PBL) is employed to promote self-learning by training thinking. The interaction between students is also an important factor. However, as is well known, traditional examinations and questionnaires only obtain subjective results. In fact, many studies have shown that brain wave data are currently the most reliable and immediate way to analyze human emotions, and are very suitable for use in evaluating things which cannot be quantified, such as the effect of learning, the appeal of music, and so on. Therefore, we boldly assume that the analysis of the brain waves can also help teachers adjust their teaching policy. Currently, most works on the analysis of brain waves, according to the rule of thumb, is to define the policy of using classification algorithms. However, the composition of human emotions is quite complex. Psychologists believe that human emotions are developed on a foundation of several basic emotions. It means that raw data on brain waves must be refined to obtain an accurate understanding of emotions. Therefore, we must focus on the degrees of classification and classification itself to find the trend of each emotion. Since living environments and cultures differ, clustering algorithms should be considered in seeking to improve the accuracy of classification. We have also developed a similarity discovery model, combined with the K-means algorithm, in proposing a more accurate framework for teaching evaluation. Our system can produce each student’s KPI. Peer rating can also establish standards. Teachers can learn about the student’s learning situations through PBL through our system, including competition among peers, the effectiveness of group discussions, active learning, and so on.