Uniform Turnpike Theorems for Finite Markov Decision Processes.
Mark E. LewisAnand A. PaulPublished in: Math. Oper. Res. (2019)
Keyphrases
- markov decision processes
- state and action spaces
- state space
- finite state
- reinforcement learning
- dynamic programming
- policy iteration
- action space
- optimal policy
- decision theoretic planning
- reachability analysis
- partially observable
- planning under uncertainty
- average reward
- markov decision process
- average cost
- transition matrices
- reinforcement learning algorithms
- infinite horizon
- action sets
- stationary policies
- model based reinforcement learning
- factored mdps
- finite horizon
- decision processes
- interval estimation
- markov decision problems
- reward function
- least squares
- continuous state spaces
- linear programming
- real time dynamic programming
- search space