New error bounds for approximations from projected linear equations

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

We consider linear fixed point equations and their approximations by projection on a low dimensional subspace. We derive new bounds on the approximation error of the solution, which are expressed in terms of low dimensional matrices and can be computed by simulation. When the fixed point mapping is a contraction, as is typically the case in Markovian decision processes (MDP), one of our bounds is always sharper than the standard worst case bounds, and another one is often sharper. Our bounds also apply to the non-contraction case, including policy evaluation in MDP with nonstandard projections that enhance exploration. There are no error bounds currently available for this case to our knowledge.

Original languageEnglish (US)
Title of host publicationRecent Advances in Reinforcement Learning - 8th European Workshop, EWRL 2008, Revised and Selected Papers
Pages253-267
Number of pages15
DOIs
StatePublished - 2008
Externally publishedYes
Event8th European Workshop on Reinforcement Learning, EWRL 2008 - Villeneuve d'Ascq, France
Duration: Jun 30 2008Jul 3 2008

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume5323 LNAI
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference8th European Workshop on Reinforcement Learning, EWRL 2008
CountryFrance
CityVilleneuve d'Ascq
Period6/30/087/3/08

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)

Fingerprint Dive into the research topics of 'New error bounds for approximations from projected linear equations'. Together they form a unique fingerprint.

Cite this