Estimating Learner’s Engagement based on Multimodal Information

Abstract

With a goal of supporting learners' self-study and improving their motivation to learn, this study proposes a method for estimating learner's engagement. First, we collected learner’s face videos, EEG, eye tracking data, and writing data. Then, we created a support vector regression (SVR) models for estimating the learner’s engagement from facial motion and EEG. We created SVR models for self-assessment, third-party assessment, and the composite score of these two measures for different learning phases: during learning, exercise, and explanation of a problem.