Game ai

[peter1591]

Q learning + deep neural network

https://medium.com/emergent-future/simple-reinforcement-learning-with-tensorflow-part-0-q-learning-with-tables-and-neural-networks-d195264329d0

[peter1591]

Multiple tree MCTS: https://arxiv.org/pdf/1102.1580.pdf

• memory efficiency?
• MO-ISMCTS: http://ieeexplore.ieee.org/abstract/document/6203567/

Belief states in MCTS http://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=7317917

[peter1591]

exploitation & exploration coefficient

• 0.7 in this paper: http://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=6203567

[peter1591]

MO-ISMCTS

• Multiple-observable information-set MCTS
• Two trees
• The actions done by opponent in their turn
  • Are edges of their tree
  • And, are projected to edges on our tree

[peter1591]

The MCTS part is now finished. Move the simulation policy to issue #55