Learning-style recognition from eye-hand movement using a dynamic Bayesian network

Educators and psychologists have debated on the efficacy of personalizing teaching methods according to learning style [1-3]. Recently, the styles are obtained from learning models, and it has been shown that utilizing those styles leverages educational effect. However, the researcher uses only two features which are the classification accuracy of human and response time, and it has been suggested that these two simple features are insufficient to characterize the complex learning styles of human [4,5]. In this work, we propose two additional features that help dramatically improve characterizing the discriminating property of people having different learning styles.