Black-box Off-policy Estimation for Infinite-Horizon Reinforcement Learning.
Ali MousaviLihong LiQiang LiuDenny ZhouPublished in: CoRR (2020)
Keyphrases
- black box
- infinite horizon
- optimal policy
- reinforcement learning
- markov decision processes
- optimal control
- partially observable
- markov decision process
- dynamic programming
- finite horizon
- state space
- long run
- stochastic demand
- policy iteration
- black boxes
- single item
- white box
- average cost
- production planning
- test cases
- lead time
- function approximation
- decision problems
- fixed cost
- markov decision problems
- reinforcement learning algorithms
- finite state
- average reward
- lost sales
- white box testing
- reward function
- temporal difference
- model free
- dec pomdps
- action selection
- markov chain