Efficiently Breaking the Curse of Horizon in Off-Policy Evaluation with Double Reinforcement Learning.
Nathan KallusMasatoshi UeharaPublished in: Oper. Res. (2022)
Keyphrases
- policy evaluation
- reinforcement learning
- temporal difference
- least squares
- model free
- function approximation
- markov decision processes
- policy iteration
- monte carlo
- variance reduction
- td learning
- state space
- high dimensional
- reinforcement learning algorithms
- optimal control
- evaluation function
- learning algorithm
- optimal policy
- semi parametric
- multi agent
- transfer learning
- neural network
- high dimensional data
- dimensionality reduction
- cost function
- machine learning