A Graphical Approach to State Variable Selection in Off-policy Learning

Joakim Blach Andersen, Qingyuan Zhao

Sequential decision problems are widely studied across many areas of science. A key challenge when learning policies from historical data - a practice commonly referred to as off-p…