Probabilistic Inference for Fast Learning in Control
File(s)
Accepted version
Author(s)
Rasmussen, Carl E
Deisenroth, Marc P
Type
Chapter
Abstract
We provide a novel framework for very fast model-based reinforcement learning in continuous state and action spaces. The framework requires probabilistic models that explicitly characterize their levels of confidence. Within this framework, we use flexible, non-parametric models to describe the world based on previously collected experience. We demonstrate learning on the cart-pole problem in a setting where we provide very limited prior knowledge about the task. Learning progresses rapidly, and a good policy is found after only a hand-full of iterations.
Editor(s)
Girgin, S
Loth, M
Munos, R
Preux, P
Ryabko, D
Date Issued
2008-11
Citation
Recent Advances in Reinforcement Learning, 2008, 5323, pp.229-242
ISBN
978-3-540-89721-7
Publisher
Springer-Verlag
Start Page
229
End Page
242
Journal / Book Title
Recent Advances in Reinforcement Learning
Lecture Notes in Computer Science
Volume
5323
Copyright Statement
© 2008 Springer-Verlag. The final publication is available at link.springer.com
Description
22.10.13 KB. Ok to add author version to spiral