Uncertainty Weighted Actor-Critic for Offline Reinforcement Learning.
Yue WuShuangfei ZhaiNitish SrivastavaJoshua M. SusskindJian ZhangRuslan SalakhutdinovHanlin GohPublished in: CoRR (2021)
Keyphrases
- actor critic
- reinforcement learning
- temporal difference
- policy gradient
- approximate dynamic programming
- reinforcement learning algorithms
- optimal control
- neuro fuzzy
- function approximation
- policy iteration
- gradient method
- model free
- dynamic programming
- state space
- learning algorithm
- temporal difference learning
- dynamical systems
- markov decision processes
- rl algorithms
- optimal policy
- policy gradient methods
- average reward
- supervised learning
- step size
- recurrent neural networks
- linear program
- learning agent
- function approximators
- partially observable
- evaluation function
- multi agent
- natural actor critic
- transfer learning