A Data-Efficient Reinforcement Learning Method Based on Local Koopman Operators.
Lixing SongJunheng WangJunhong XuPublished in: ICMLA (2021)
Keyphrases
- synthetic data
- data sets
- prior knowledge
- noisy data
- input data
- raw data
- statistical methods
- test data
- dynamic programming
- computationally efficient
- training samples
- missing data
- database
- statistical analysis
- significant improvement
- reinforcement learning
- training data
- data sources
- detection method
- data analysis
- information loss
- high accuracy
- correlation analysis
- clustering method
- data structure
- segmentation method
- control policy
- neural network
- prior information
- original data
- high quality
- missing values
- markov decision processes
- privacy preserving
- knowledge discovery
- data collection