Uncertainty Weighted Actor-Critic for Offline Reinforcement Learning.
Yue WuShuangfei ZhaiNitish SrivastavaJoshua M. SusskindJian ZhangRuslan SalakhutdinovHanlin GohPublished in: ICML (2021)
Keyphrases
- actor critic
- reinforcement learning
- temporal difference
- policy gradient
- reinforcement learning algorithms
- optimal control
- approximate dynamic programming
- neuro fuzzy
- gradient method
- function approximation
- control problems
- optimal policy
- model free
- state space
- machine learning
- policy iteration
- policy gradient methods
- markov decision processes
- learning algorithm
- learning problems
- average reward
- rl algorithms
- natural actor critic
- reinforcement learning methods
- evaluation function
- transfer learning
- monte carlo
- dynamic environments
- dynamic programming
- search space
- multi agent