l1-penalized projected Bellman residual - Inria - Institut national de recherche en sciences et technologies du numérique Accéder directement au contenu
Communication Dans Un Congrès Année : 2011

l1-penalized projected Bellman residual

Matthieu Geist
Bruno Scherrer

Résumé

We consider the task of feature selection for value function approximation in reinforcement learning. A promising approach consists in combining the Least-Squares Temporal Difference (LSTD) algorithm with $\ell_1$-regularization, which has proven to be effective in the supervised learning community. This has been done recently whit the LARS-TD algorithm, which replaces the projection operator of LSTD with an $\ell_1$-penalized projection and solves the corresponding fixed-point problem. However, this approach is not guaranteed to be correct in the general off-policy setting. We take a different route by adding an $\ell_1$-penalty term to the projected Bellman residual, which requires weaker assumptions while offering a comparable performance. However, this comes at the cost of a higher computational complexity if only a part of the regularization path is computed. Nevertheless, our approach ends up to a supervised learning problem, which let envision easy extensions to other penalties.
Fichier principal
Vignette du fichier
gs_ewrl_l1_cr.pdf (236.6 Ko) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)
Loading...

Dates et versions

hal-00644507 , version 1 (24-11-2011)

Identifiants

  • HAL Id : hal-00644507 , version 1

Citer

Matthieu Geist, Bruno Scherrer. l1-penalized projected Bellman residual. European Wrokshop on Reinforcement Learning (EWRL 11), Sep 2011, Athens, Greece. ⟨hal-00644507⟩
195 Consultations
468 Téléchargements

Partager

Gmail Facebook X LinkedIn More