Challenges and Opportunities in Offline Reinforcement Learning from Visual Observations.
Cong LuPhilip J. BallTim G. J. RudnerJack Parker-HolderMichael A. OsborneYee Whye TehPublished in: CoRR (2022)
Keyphrases
- reinforcement learning
- lessons learned
- function approximation
- low level
- visual perception
- visual information
- real time
- data mining
- high level
- open issues
- genetic algorithm
- real world
- multi agent reinforcement learning
- policy search
- temporal difference learning
- application scenarios
- technical challenges
- visual cues
- visual data
- markov decision processes
- visual features
- dynamic programming
- learning process
- information retrieval