My great thanks to Chris Fregly and the Hadoop Summit for inviting me to speak at the Advanced Spark and Tensor Flow meetup last night. It was a great pleasure meeting everyone and talking about the deep reinforcement learning work we are doing at foodRev.
There are two papers I mentioned while talking with people afterwards that are super great:
Bandit based Monte-Carlo Planning (referenced by the AlphaGo paper), which references another super good paper - A Sparse Sampling Algorithm for Near-Optimal Planning in Large Markov Decision Processes from Andrew Kearns, Yishay Mansour, and the ubiquitous Andrew Ng.