Optimal Adaptive Policies for Markov Decision Processes.
Apostolos N. BurnetasMichael N. KatehakisPublished in: Math. Oper. Res. (1997)
Keyphrases
- markov decision processes
- optimal policy
- average cost
- stationary policies
- dynamic programming
- discounted reward
- total reward
- average reward
- policy iteration algorithm
- finite horizon
- state space
- markov decision process
- discount factor
- finite state
- action sets
- expected reward
- reinforcement learning
- decision processes
- control policies
- decision theoretic planning
- reward function
- partially observable
- action space
- transition matrices
- policy iteration
- reinforcement learning algorithms
- macro actions
- decision problems
- optimality criterion
- decentralized control
- partially observable markov decision processes
- optimal control
- infinite horizon
- reachability analysis
- long run
- risk sensitive
- factored mdps
- markov decision problems
- control policy
- model based reinforcement learning
- multistage
- lot sizing
- state abstraction
- linear program
- state and action spaces
- initial state
- real time dynamic programming