Novel Feature Selection and Kernel-Based Value Approximation Method for Reinforcement Learning.
Hunor Sandor JakabLehel CsatóPublished in: ICANN (2013)
Keyphrases
- feature selection
- preprocessing
- reinforcement learning
- dynamic programming
- method for feature selection
- unsupervised learning
- mutual information
- feature set
- high accuracy
- objective function
- computational complexity
- data sets
- classification method
- text categorization
- machine learning
- learning algorithm
- face recognition
- classification accuracy
- decision trees
- model selection
- similarity measure
- segmentation method
- cross validation
- significant improvement
- transition model
- pairwise