Particle Filter-based Policy Gradient in POMDPs.
Pierre-Arnaud CoquelinRomain DeguestRémi MunosPublished in: NIPS (2008)
Keyphrases
- policy gradient
- partially observable markov decision processes
- reinforcement learning
- gradient ascent
- actor critic
- gradient method
- function approximation
- optimal control
- policy gradient methods
- particle filter
- object tracking
- policy search
- reinforcement learning algorithms
- variance reduction
- approximation methods
- model free reinforcement learning
- average reward
- reinforcement learning methods
- markov decision processes
- sufficient conditions