Learning to Discover Task-Relevant Features for Interpretable Reinforcement Learning.
Qiyuan ZhangXiaoteng MaYiqin YangChenghao LiJun YangYu LiuBin LiangPublished in: IEEE Robotics Autom. Lett. (2021)
Keyphrases
- reinforcement learning
- learning process
- prior knowledge
- action selection
- learning tasks
- multi agent reinforcement learning
- reinforcement learning methods
- learning problems
- learning algorithm
- function approximators
- feature set
- active exploration
- highly informative
- feature construction
- background knowledge
- learning systems
- knowledge acquisition
- online learning
- co occurrence
- supervised learning
- image features
- active learning
- feature vectors