Robustness of policies in constrained Markov decision processes.
Alexander ZadorojniyAdam ShwartzPublished in: IEEE Trans. Autom. Control. (2006)
Keyphrases
- markov decision processes
- optimal policy
- markov decision process
- decision processes
- average cost
- state space
- reward function
- finite state
- total reward
- dynamic programming
- policy iteration
- decentralized control
- reinforcement learning
- average reward
- decision problems
- macro actions
- discounted reward
- transition matrices
- policy iteration algorithm
- finite horizon
- reachability analysis
- reinforcement learning algorithms
- control policies
- stationary policies
- model based reinforcement learning
- infinite horizon
- markov decision problems
- partially observable
- partially observable markov decision processes
- multistage
- planning under uncertainty
- factored mdps
- long run
- decision theoretic planning
- semi markov decision processes
- expected reward
- action space
- action sets
- risk sensitive
- optimal control
- state and action spaces
- probabilistic planning
- multi agent