Information Directed Reward Learning for Reinforcement Learning.
David LindnerMatteo TurchettaSebastian TschiatschekKamil CiosekAndreas KrausePublished in: NeurIPS (2021)
Keyphrases
- reinforcement learning
- learning algorithm
- learning process
- information sources
- end users
- function approximation
- prior knowledge
- information extraction
- supervised learning
- online learning
- eligibility traces
- dynamic programming
- state space
- user interaction
- markov decision processes
- reinforcement learning algorithms
- robot control
- learning agents
- temporal difference learning
- active exploration