Statistical Reinforcement Learning at Michigan

We generalize existing and develop new algorithms and methods in Reinforcement Learning for use in improving health and well-being. We are particularly interesting in combining statistical methods for conducting inference (confidence intervals, hypothesis tests) with algorithmic methods developed in computer science for use in learning and evaluating treatment policies (dynamic treatment regimes). To learn more about dynamic treatment regimes see the wikipedia page.

Lab Members

Research Papers


Software for SMART Studies can be found here

The Methodology Center at PennState

Internal Site

Susan A. Murphy,
Department of Statistics & Institute for Social Research
University of Michigan
Ann Arbor, MI

[ back to top page ]