Show simple item record

Autonomous Robots

dc.contributor.authorSugiyama, Masashi
dc.contributor.authorHachiya, Hirotaka
dc.contributor.authorTowell, Christopher
dc.contributor.authorVijayakumar, Sethu
dc.date.accessioned2010-08-31T13:49:26Z
dc.date.available2010-08-31T13:49:26Z
dc.date.issued2008
dc.identifier.urihttp://www.springerlink.com/content/4j2g52m1272hj185/en
dc.identifier.urihttp://hdl.handle.net/1842/3697
dc.description.abstractThe least-squares policy iteration approach works efficiently in value function approximation, given appropriate basis functions. Because of its smoothness, the Gaussian kernel is a popular and useful choice as a basis function. However, it does not allow for discontinuity which typically arises in real-world reinforcement learning tasks. In this paper, we propose a new basis function based on geodesic Gaussian kernels, which exploits the non-linear manifold structure induced by the Markov decision processes. The usefulness of the proposed method is successfully demonstrated in simulated robot arm control and Khepera robot navigation.en
dc.language.isoenen
dc.publisherSpringeren
dc.subjectReinforcement learningen
dc.subjectValue function approximationen
dc.subjectMarkov decision processen
dc.subjectLeast-squares policy iterationen
dc.subjectGaussian kernelen
dc.titleGeodesic Gaussian kernels for value function approximationen
dc.typeArticleen
dc.identifier.doi10.1007/s10514-008-9095-6en
rps.issue3en
rps.volume25en
rps.titleAutonomous Robotsen
dc.extent.pageNumbers287-304en
dc.date.updated2010-08-31T13:49:26Z
dc.identifier.eIssn09295593en


Files in this item

This item appears in the following Collection(s)

Show simple item record