Model-free Reinforcement Learning in Infinite-horizon Average-reward Markov Decision Processes.
Chen-Yu WeiMehdi Jafarnia-JahromiHaipeng LuoHiteshi SharmaRahul JainPublished in: CoRR (2019)
Keyphrases
- infinite horizon
- average reward
- markov decision processes
- policy gradient
- optimal policy
- finite horizon
- policy iteration
- long run
- reinforcement learning
- finite state
- state space
- stochastic games
- dynamic programming
- state action
- partially observable
- discounted reward
- average cost
- planning under uncertainty
- markov decision process
- discount factor
- reinforcement learning algorithms
- partially observable markov decision processes
- multistage
- total reward
- reward function
- optimal control
- search algorithm