A new decomposition technique for solving Markov decision processes.
Pierre LarocheYann BonifaceRené SchottPublished in: SAC (2001)
Keyphrases
- markov decision processes
- transition matrices
- semi markov decision processes
- optimal policy
- policy iteration
- state space
- finite state
- reinforcement learning
- planning under uncertainty
- average reward
- markov decision problems
- dynamic programming
- stochastic shortest path
- reachability analysis
- factored mdps
- decision theoretic planning
- finite horizon
- average cost
- infinite horizon
- action space
- risk sensitive
- model based reinforcement learning
- policy iteration algorithm
- decision processes
- reinforcement learning algorithms
- state and action spaces
- reward function
- linear programming
- state abstraction
- markov decision process
- discounted reward