Maximum Entropy Reinforcement Learning in Two-Player Perfect Information Games.
Taichi NakayashikiTomoyuki KanekoPublished in: SSCI (2021)
Keyphrases
- perfect information
- maximum entropy
- reinforcement learning
- imperfect information
- card games
- maximum entropy principle
- subgame perfect equilibrium
- card game
- board game
- state space
- temporal difference
- minimum cross entropy
- reinforcement learning algorithms
- human players
- optimal policy
- learning algorithm
- model free
- machine learning
- game tree
- conditional random fields
- dynamic programming
- game playing
- learning agents
- evaluation function
- markov decision processes
- optimal solution