Average Cost Optimality Inequality for Markov Decision Processes with Borel Spaces and Universally Measurable Policies.
Huizhen YuPublished in: SIAM J. Control. Optim. (2020)
Keyphrases
- markov decision processes
- average cost
- optimal policy
- stationary policies
- finite horizon
- finite state
- policy iteration
- state space
- reinforcement learning
- markov decision process
- action sets
- holding cost
- approximate dynamic programming
- control policy
- risk sensitive
- dynamic programming
- transition matrices
- decision theoretic planning
- infinite horizon
- reinforcement learning algorithms
- initial state
- markov decision problems
- long run
- reward function
- decision processes
- decision problems
- average reward
- total reward
- multistage
- markov chain
- partially observable
- function approximation
- state dependent
- action space
- random variables
- partially observable markov decision processes