Goal-driven dimensionality reduction for reinforcement learning.
Simone ParisiSimon RamstedtJan PetersPublished in: IROS (2017)
Keyphrases
- goal driven
- dimensionality reduction
- reinforcement learning
- software product line
- low dimensional
- service composition
- function approximation
- principal component analysis
- high dimensional
- high dimensional data
- pattern recognition
- feature extraction
- state space
- semantic web services
- high dimensionality
- optimal policy
- learning algorithm
- principal components
- data points
- markov decision processes
- machine learning
- databases
- database
- future directions
- learning process
- feature space
- feature selection
- kernel learning