In-Sample Policy Iteration for Offline Reinforcement Learning.
Xiaohan HuYi MaChenjun XiaoYan ZhengZhaopeng MengPublished in: CoRR (2023)
Keyphrases
- policy iteration
- reinforcement learning
- markov decision processes
- model free
- optimal policy
- temporal difference
- sample path
- policy evaluation
- markov decision process
- stochastic approximation
- fixed point
- average reward
- least squares
- reinforcement learning algorithms
- function approximation
- approximate dynamic programming
- temporal difference learning
- actor critic
- state space
- infinite horizon
- learning algorithm
- optimal control
- sample size
- approximate policy iteration
- markov decision problems
- partially observable
- finite state
- machine learning
- convergence rate
- control problems
- multistage
- monte carlo
- dynamic programming
- multi agent
- evaluation function
- random walk
- linear programming
- supervised learning
- semi supervised
- objective function