Partially Observable Risk-Sensitive Markov Decision Processes.
Nicole BäuerleUlrich RiederPublished in: Math. Oper. Res. (2017)
Keyphrases
- markov decision processes
- risk sensitive
- partially observable
- markov decision problems
- state space
- optimal policy
- dynamic programming
- finite state
- reinforcement learning
- average cost
- infinite horizon
- planning under uncertainty
- reward function
- finite horizon
- decision processes
- reinforcement learning algorithms
- policy iteration
- average reward
- markov decision process
- action space
- markov chain
- control policies
- multistage
- sufficient conditions