Talk
"Towards Decision-Theoretic Teaching: Sequential Planning Under Uncertainty in Domains Inspired By Tutoring"
Emma Brunskill, UC Berkeley
Tuesday, August 3, 2010 at 12:00 Noon
Room 368 (CIT 3rd Floor)
I will present ongoing work that uses and advances research on sequential decision making in partially observable environments towards challenges in automated teaching. First, prior education work suggests student knowledge may be coarsely captured by factored, topological models. We leverage this structure in an anytime planner for such domains. Second, motivated by the need for multi-user interfaces in settings with limited computers, we have recently used a sequential decision making under uncertainty approach to develop an adaptive interface for a multi-user mathematical drill exercise game. Preliminary results from a controlled trial with 2 schools in Bangalore, India highlight the potential of this approach. Finally I will briefly discuss our work on exploring various models of student learning as part of a concept learning (psychology) task. Throughout the talk I will highlight the technical and theoretical challenges posed by these problems, as well as the potential practical benefit of posing automated teaching as a sequential decision making problem.
Emma Brunskill is a NSF Mathematical Sciences Postdoctoral fellow at UC Berkeley. She received a BS in Computer Engineering and Physics from the University of Washington, a MS in Neuroscience from Oxford University as a Rhodes Scholar, and a PhD in Computer Science from MIT. Her research interests include reinforcement learning, decision making under uncertainty, and using information communication technologies for international development.
This is joint work with Stuart Russell, Leah Findlater, Sunil Garg, Clint Tseng, Joyojeet Pal, Anna Rafferty, Pat Shafto and Tom Griffiths.
Hosts: Amy Greenwald and Erik Sudderth