IC13 – Personalizing instruction and recognizing student misunderstandings using reinforcement learning

Lecturer: Anna Rafferty
Fields: Artifical Intelligence/Machine learning


Online educational technologies provide opportunities to monitor learners’ knowledge in real time and modify instruction based on learners’ responses. In this talk, I’ll give a brief overview of some of the ways that reinforcement learning has been used to achieve these goals, and then provide a more in-depth discussion of my own work using inverse reinforcement learning to make inferences about learners’ understanding. In this work, we are particularly focused on interpreting learners’ behavior in multi-step tasks, such as games or mathematical problem solving, and we combine ideas from machine learning and computational cognitive modeling. Our approach offers the potential to provide feedback about learners’ strategies and misunderstandings based on their pattern of interactions. Overall, the talk will argue that work in reinforcement learning for education has the potential to create smarter educational resources and that taking an interdisciplinary perspective suggests new insights and approaches.


  • Rafferty, A. N., Jansen R. A., & Griffiths, T. L. (2020). Assessing Mathematics Misunderstandings via Bayesian Inverse Planning. Cognitive Science. DOI: 10.1111/cogs.12900
  • Rafferty, A. N., Jansen, R. A., & Griffiths, T. L. (2016) Using Inverse Planning for Personalized Feedback. Proceedings of the 9th International Conference on Educational Data Mining (pp. 472-477). http://tiny.cc/IRLFeedbackEDM2016


Anna Rafferty

Dr. Anna Rafferty earned her PhD from the University of California, Berkeley, and is currently an associate professor of computer science at Carleton College. Her work addresses questions at the intersection of machine learning, computational cognitive science, and education. She is particularly interested in developing automated strategies to provide effective feedback to students and in developing technologies that can both continuously improve instruction for students and provide valuable data for researchers to draw more general conclusions about the effectiveness of educational interventions. Dr. Rafferty has recently begun work emphasizing the importance of considering equitable impacts across students in educational technologies that have the potential for personalization.

Affiliation: Carleton College
Homepage: https://sites.google.com/site/annanrafferty/