Pure Stationary Optimal Strategies in Markov Decision Processes.
Hugo GimbertPublished in: STACS (2007)
Keyphrases
- markov decision processes
- optimal strategy
- decision problems
- optimal policy
- stationary policies
- state space
- non stationary
- monte carlo
- finite state
- expected cost
- reinforcement learning
- dynamic programming
- policy iteration
- transition matrices
- partially observable
- expected utility
- infinite horizon
- factored mdps
- decision theoretic planning
- mathematical models
- reachability analysis
- planning under uncertainty
- influence diagrams
- finite horizon
- action space
- artificial intelligence
- average reward
- average cost
- reward function
- reinforcement learning algorithms
- state abstraction
- semi markov decision processes
- model based reinforcement learning
- markov decision process
- initial state
- state and action spaces
- action sets