Asymptotically Exact Error Characterization of Offline Policy Evaluation with Misspecified Linear Models.
Kohei MiyaguchiPublished in: NeurIPS (2021)
Keyphrases
- linear models
- policy evaluation
- least squares
- variance reduction
- linear model
- linear regression
- variable selection
- temporal difference
- sample size
- reinforcement learning
- monte carlo
- model free
- markov decision processes
- function approximation
- gaussian processes
- policy iteration
- optimal policy
- semi parametric
- cross validation
- machine learning
- causal relationships
- worst case