Predictive representations for policy gradient in POMDPs.
Abdeslam BoulariasBrahim Chaib-draaPublished in: ICML (2009)
Keyphrases
- policy gradient
- reinforcement learning
- actor critic
- gradient ascent
- function approximation
- policy gradient methods
- partially observable markov decision processes
- policy search
- gradient method
- variance reduction
- reinforcement learning algorithms
- model free reinforcement learning
- optimal control
- approximation methods
- temporal difference
- markov decision processes
- single agent
- markov chain
- state space
- dynamic programming
- neural network