Model-Selection for Non-parametric Function Approximation in Continuous Control Problems: A Case Study in a Smart Energy System.
Daniel UrieliPeter StonePublished in: ECML/PKDD (1) (2013)
Keyphrases
- model selection
- function approximation
- control problems
- reinforcement learning
- gaussian process
- cross validation
- machine learning
- hyperparameters
- optimal control
- action space
- temporal difference
- temporal difference learning
- sample size
- model free
- reinforcement learning algorithms
- adaptive control
- function approximators
- model selection criteria
- state space
- optimal policy
- feature selection
- learning algorithm
- radial basis function
- regression model
- markov decision processes
- dynamic programming
- statistical inference
- transfer learning
- action selection
- learning process
- data sets
- continuous state
- reinforcement learning methods
- training data
- learning tasks