A Theoretical Understanding of Gradient Bias in Meta-Reinforcement Learning.
Bo LiuXidong FengJie RenLuo MaiRui ZhuHaifeng ZhangJun WangYaodong YangPublished in: NeurIPS (2022)
Keyphrases
- reinforcement learning
- theoretical analysis
- neural network
- policy gradient
- deeper understanding
- real valued
- edge detection
- state space
- transfer learning
- learning process
- optimal control
- temporal difference
- machine learning
- action space
- data sets
- multi agent reinforcement learning
- policy search
- theoretical and practical implications