Approximate Policy Iteration with a Policy Language Bias: Solving Relational Markov Decision Processes.
Alan FernSung Wook YoonRobert GivanPublished in: J. Artif. Intell. Res. (2006)
Keyphrases
- markov decision processes
- approximate policy iteration
- policy iteration
- markov decision problems
- markov games
- transition matrices
- optimal policy
- partially observable
- state space
- reinforcement learning
- markov decision process
- finite state
- average cost
- infinite horizon
- average reward
- decision processes
- linear programming
- finite horizon
- dynamic programming
- reward function
- least squares
- action space
- reinforcement learning algorithms
- temporal difference
- policy search
- planning under uncertainty
- decision problems
- transition probabilities
- utility function
- sufficient conditions
- model free
- control policies
- evaluation function
- learning tasks
- multi agent
- machine learning