Approximate discounting-free policy evaluation from transient and recurrent states.
Vektor DewantoMarcus GallagherPublished in: CoRR (2022)
Keyphrases
- policy evaluation
- least squares
- monte carlo
- temporal difference
- reinforcement learning
- model free
- policy iteration
- variance reduction
- markov decision processes
- markov decision problems
- function approximation
- semi parametric
- steady state
- optimal policy
- step size
- gaussian process
- statistical inference
- initial state
- action selection
- state transitions
- policy gradient
- decision making
- partially observable markov decision processes
- importance sampling
- state space
- finite state
- evaluation function
- belief revision
- fixed point
- markov chain