SVRG for Policy Evaluation with Fewer Gradient Evaluations.
Zilun PengAhmed TouatiPascal VincentDoina PrecupPublished in: IJCAI (2020)
Keyphrases
- policy evaluation
- least squares
- temporal difference
- reinforcement learning
- monte carlo
- model free
- policy iteration
- matrix inversion
- variance reduction
- policy gradient
- function approximation
- markov decision processes
- semi parametric
- partially observable markov decision processes
- fixed point
- optimal policy
- gaussian process
- finite state
- linear regression
- gradient method
- markov decision problems
- state space