ORBi: Fonteneau Raphaël - Inferring bounds on the performance of a control policy from a sample of trajectories

Reference : Inferring bounds on the performance of a control policy from a sample of trajectories


	Document type :	Scientific congresses and symposiums : Paper published in a book
	Discipline(s) :	Engineering, computing & technology : Computer science
	To cite this reference:	http://hdl.handle.net/2268/13667

Title :	Inferring bounds on the performance of a control policy from a sample of trajectories
Language :	English
Author, co-author :	Fonteneau, Raphaël [Université de Liège - ULg > Dép. d'électric., électron. et informat. (Inst.Montefiore) > Systèmes et modélisation >]
	Murphy, Susan [ > > ]
	Wehenkel, Louis [Université de Liège - ULg > Dép. d'électric., électron. et informat. (Inst.Montefiore) > Systèmes et modélisation >]
	Ernst, Damien [Université de Liège - ULg > Dép. d'électric., électron. et informat. (Inst.Montefiore) > Systèmes et modélisation >]
Publication date :	2009
Main document title :	Proceedings of the IEEE International Symposium on Adaptive Dynamic Programming and Reinforcement Learning (ADPRL-09)
Pages :	117-123
Peer reviewed :	Yes
On invitation :	No
Audience :	International
ISBN :	978-1-4244-2761-1
Event name :	IEEE International Symposium on Adaptive Dynamic Programming and Reinforcement Learning (ADPRL-09)
Event date :	March 30 - April 2, 2009
Event place (city) :	Nashville
Event country :	USA
Keywords :	[en] reinforcement learning ; model-free ; lower bound on a policy
	[fr] performance guarantee
Abstract :	[en] We propose an approach for inferring bounds on the finite-horizon return of a control policy from an off-policy sample of trajectories collecting state transitions, rewards, and control actions. In this paper, the dynamics, control policy, and reward function are supposed to be deterministic and Lipschitz continuous. Under these assumptions, a polynomial algorithm, in terms of the sample size and length of the optimization horizon, is derived to compute these bounds, and their tightness is characterized in terms of the sample density.
Funders :	Fonds de la Recherche Scientifique (Communauté française de Belgique) - F.R.S.-FNRS
Permalink :	http://hdl.handle.net/2268/13667
DOI :	10.1109/ADPRL.2009.4927534

File(s) associated to this reference

Fulltext file(s):

	File	Commentary	Version	Size	Access
Open access	bounds-trajectories-adprl.pdf		Publisher postprint	249.44 kB	View/Open

Additional material(s):

	File	Commentary	Version	Size	Access
Open access	RL-TU-Delft-2009.pdf		-	292.91 kB	View/Open

All documents in ORBi are protected by a user license.

Dspace v1.5