Fitted Q-iteration in continuous action-space MDPs - Inria - Institut national de recherche en sciences et technologies du numérique Access content directly
Reports (Technical Report) Year : 2007

Fitted Q-iteration in continuous action-space MDPs

Rémi Munos
  • Function : Author
  • PersonId : 836863
Csaba Szepesvari
  • Function : Author
  • PersonId : 844057


We consider continuous state, continuous action batch reinforcement learning where the goal is to learn a good policy from a sufficiently rich trajectory generated by some policy. We study a variant of fitted Q-iteration, where the greedy action selection is replaced by searching for a policy in a restricted set of candidate policies by maximizing the average action values. We provide a rigorous analysis of this algorithm, proving what we believe is the first finite-time bound for value-function based algorithms for continuous state and action problems.
Fichier principal
Vignette du fichier
rlca.pdf (370.43 Ko) Télécharger le fichier
Origin : Files produced by the author(s)

Dates and versions

inria-00185311 , version 1 (05-11-2007)
inria-00185311 , version 2 (08-01-2008)


  • HAL Id : inria-00185311 , version 2


Andras Antos, Rémi Munos, Csaba Szepesvari. Fitted Q-iteration in continuous action-space MDPs. [Technical Report] 2007, pp.24. ⟨inria-00185311v2⟩
456 View
775 Download


Gmail Facebook X LinkedIn More