Unifying Gradient Estimators for Meta-Reinforcement Learning via Off-Policy Evaluation.
Yunhao TangTadashi KozunoMark RowlandRémi MunosMichal ValkoPublished in: CoRR (2021)
Keyphrases
- policy evaluation
- gradient estimators
- reinforcement learning
- temporal difference
- least squares
- model free
- monte carlo
- policy iteration
- function approximation
- markov decision processes
- td learning
- variance reduction
- state space
- evaluation function
- semi parametric
- optimal policy
- statistical inference
- machine learning
- partially observable markov decision processes
- semi supervised
- dynamic programming
- multi agent
- learning algorithm
- optimal control
- neural network
- regression model
- importance sampling