Another set of verifiable conditions for average Markov decision processes with Borel spaces.
Xiaolong ZouXianping GuoPublished in: Kybernetika (2015)
Keyphrases
- markov decision processes
- state space
- finite state
- optimal policy
- average cost
- stationary policies
- reinforcement learning
- dynamic programming
- sufficient conditions
- discounted reward
- planning under uncertainty
- transition matrices
- reachability analysis
- decision theoretic planning
- policy iteration
- average reward
- risk sensitive
- factored mdps
- probability distribution
- machine learning
- decision processes
- finite horizon
- partially observable
- linear program