Computationally Feasible Bounds for Partially Observed Markov Decision Processes.
William S. LovejoyPublished in: Oper. Res. (1991)
Keyphrases
- computationally feasible
- markov decision processes
- partially observed
- discounted reward
- optimal policy
- transition matrices
- lower bound
- finite state
- reinforcement learning
- state space
- decision theoretic planning
- dynamic programming
- planning under uncertainty
- exhaustive search
- reachability analysis
- policy iteration
- model based reinforcement learning
- state and action spaces
- factored mdps
- decision processes
- infinite horizon
- risk sensitive
- markov decision process
- upper and lower bounds
- average reward
- action space
- average cost
- partially observable
- finite horizon
- reinforcement learning algorithms
- neural network
- expected reward
- evolutionary algorithm