In this presentation, Dr. Jonathan Dodge presents Mastering the game of Stratego with model-free multiagent reinforcement learning, by Perolat et al. In it, the authors dispose of the Monte Carlo Tree Search (MCTS) systems of previous solutions, opting instead for a game theoretic approach.
Presentation Link: https://psu.mediaspace.kaltura.com/media/Group+Meeting/1_gqevckug