Near-optimal Policy Identification in Active Reinforcement Learning.
Xiang LiViraj MehtaJohannes KirschnerIan CharWillie NeiswangerJeff SchneiderAndreas KrauseIlija BogunovicPublished in: CoRR (2022)
Keyphrases
- optimal policy
- reinforcement learning
- markov decision processes
- state space
- decision problems
- finite horizon
- dynamic programming
- finite state
- infinite horizon
- multistage
- state dependent
- average reward
- markov decision process
- long run
- function approximation
- sufficient conditions
- policy iteration
- bayesian reinforcement learning
- markov decision problems
- partially observable markov decision processes
- initial state
- reinforcement learning algorithms
- average cost
- model free
- partially observable
- reward function
- control policies
- stochastic demand
- dynamical systems
- serial inventory systems