Hybrid Least-Squares Algorithms for Approximate Policy Evaluation.
Jeffrey JohnsMarek PetrikSridhar MahadevanPublished in: ECML/PKDD (1) (2009)
Keyphrases
- policy evaluation
- least squares
- hybrid algorithms
- policy iteration
- reinforcement learning
- monte carlo
- model free
- matrix inversion
- markov decision processes
- temporal difference
- semi parametric
- linear regression
- variance reduction
- function approximation
- learning algorithm
- linear model
- statistical inference
- artificial neural networks
- decision trees