Bayesian active learning-based robot tutor for children's word-reading skills

摘要: Effective tutoring requires personalization of the interaction to each student. Continuous and efficient assessment student's skills are a prerequisite for such personalization. We developed Bayesian active-learning algorithm that continuously efficiently assesses child's word-reading implemented it in social robot. then an integrated experimental paradigm which child plays novel story-creation tablet game with The robot is portrayed as younger peer who wishes learn read, framing well empowering child. show our results accurate representation large age range, 4-8 year old children, initial reading skill range. also employing child-specific assessment-based age- skill-independent learning, compared random tutoring. Finally, system enables us implementing same learning on robot's knowledge comparable what thinks has learned. perception age-dependent may facilitate indirect development theory-of-mind.