Principal Investigator
Dr. Ansaf Salleb-Aouissi
Description
I am creating an open-source and state-of-the-art multi-armed bandit (MAB) intelligent tutoring algorithm to education technology literature for instructors’ use for remote education. My algorithm utilizes novel hierarchical MABs, where separate MAB agents are called to select a concept and dependently select a problem. This algorithm also takes into account predefined problem difficulties from educational supervisors, adapted from prior literature. With perform Bayesian Knowledge Tracing simulations and a controlled real trial on our entire Aiphabet AI literacy boot camp course with high school students, we compare the differences in our learning platform’s effectiveness with (a) a static, domain expert-defined progression sequence, and (b) our hierarchical MAB-defined progression sequence.
This work is adapted from Mu et al. (2018) and Segal et al. (2018) . Readers of this page are highly encouraged to explore both conference papers before my own is finalized!