Policy Iteration Based on a Learned Transition Model.
Vivek RamavajjalaCharles ElkanPublished in: ECML/PKDD (2) (2012)
Keyphrases
- transition model
- policy iteration
- markov decision problems
- reinforcement learning
- markov decision processes
- optimal policy
- model free
- least squares
- reward function
- fixed point
- infinite horizon
- markov decision process
- temporal difference
- finite state
- linear programming
- convergence rate
- state space
- partially observable
- optimal control
- dynamic programming
- function approximation
- supervised learning
- decision processes
- hidden markov models
- neural network
- linear program
- monte carlo
- sufficient conditions
- reinforcement learning algorithms
- partially observable markov decision processes
- learning algorithm