Thinking Fast and Slow with Deep Learning and Tree Search

Thomas Anthony, Zheng Tian, David Barber

Sequential decision making problems, such as structured prediction, robotic control, and game playing, require a combination of planning policies and generalisation of those plans.…