On-Line Search for Solving Markov Decision Processes via Heuristic Sampling.
Laurent PéretFrédérick GarciaPublished in: ECAI (2004)
Keyphrases
- markov decision processes
- transition matrices
- dynamic programming
- search algorithm
- semi markov decision processes
- state space
- optimal policy
- search strategies
- policy iteration
- reinforcement learning
- average reward
- heuristic function
- markov decision problems
- stochastic shortest path
- reinforcement learning algorithms
- finite state
- reachability analysis
- decision theoretic planning
- finite horizon
- infinite horizon
- average cost
- factored mdps
- search space
- partially observable
- model based reinforcement learning
- action sets
- state and action spaces
- hill climbing
- planning graph
- decision processes
- action space
- risk sensitive
- admissible heuristics
- discounted reward
- least squares
- machine learning
- reward function