Using Reinforcement Learning for Operating Educational Campuses Safely during a Pandemic (Student Abstract)

Abstract The COVID-19 pandemic has brought a significant disruption not only on how schools operate but also affected student sentiments on learning and adoption to different learning strategies. We propose CampusPandemicPlanR, a reinforcement learning-based simulation tool that could be applied to suggest to campus operators how many students from each course to allow on a campus classroom each week. The tool aims to strike a balance between the conflicting goals of keeping students from getting infected, on one hand, and allowing more students to come into campus to allow them to benefit from in-person classes, on the other. Our preliminary results show that reinforcement learning is able to learn better policies over iterations, and that different Pareto-optimal tradeoffs between these conflicting goals could be obtained by varying the reward weight parameter.