Fast Approximate Dynamic Programming for Infinite-Horizon Markov Decision Processes.
Mohamad Amin Sharifi KolarijaniG. F. MaxPeyman Mohajerin Mohajerin EsfahaniPublished in: NeurIPS (2021)
Keyphrases
- approximate dynamic programming
- infinite horizon
- markov decision processes
- policy iteration
- average cost
- dynamic programming
- reinforcement learning
- optimal policy
- finite horizon
- state space
- finite state
- average reward
- markov decision process
- factored mdps
- partially observable
- linear program
- planning under uncertainty
- long run
- decision processes
- multistage
- actor critic
- optimal control
- control policy
- initial state
- action space
- stationary policies
- state dependent
- markov decision problems
- dec pomdps
- search algorithm