Ultimately Stationary Policies to Approximate Risk-Sensitive Discounted MDPs.
Uday Kumar MSanjay P. BhatVeeraruna KavithaNandyala HemachandraPublished in: VALUETOOLS (2019)
Keyphrases
- risk sensitive
- markov decision processes
- stationary policies
- action sets
- average cost
- optimal policy
- finite state
- markov decision process
- reinforcement learning
- dynamic programming
- policy iteration
- state space
- average reward
- finite horizon
- infinite horizon
- markov decision problems
- planning under uncertainty
- decision processes
- reinforcement learning algorithms
- machine learning
- optimal control
- partially observable
- linear programming
- search space
- supervised learning
- action space
- utility function
- multistage
- objective function
- sufficient conditions