Show simple item record

dc.contributor.authorHazan, Elad
dc.contributor.authorLi, Yuanzhi
dc.date.accessioned2019-10-23T23:19:32Z
dc.date.available2019-10-23T23:19:32Z
dc.date.created2017-01-05 00:28
dc.date.issued2016-03-14
dc.identifieroai:arXiv.org:1603.04350
dc.identifierhttp://arxiv.org/abs/1603.04350
dc.identifier.urihttp://hdl.handle.net/20.500.12424/784321
dc.description.abstractWe consider the problem of online convex optimization against an arbitrary adversary with bandit feedback, known as bandit convex optimization. We give the first $\tilde{O}(\sqrt{T})$-regret algorithm for this setting based on a novel application of the ellipsoid method to online learning. This bound is known to be tight up to logarithmic factors. Our analysis is introduces new tools in discrete convex geometry.
dc.description.abstractComment: 28 pages, 7 figures
dc.subjectComputer Science - Learning
dc.subjectComputer Science - Data Structures and Algorithms
dc.subjectG.1.6
dc.titleAn optimal algorithm for bandit convex optimization
dc.typetext
ge.collectioncodeOAIDATA
ge.dataimportlabelOAI metadata object
ge.identifier.legacyglobethics:10377645
ge.identifier.permalinkhttps://www.globethics.net/gel/10377645
ge.lastmodificationdate2017-01-05 00:28
ge.lastmodificationuseradmin@pointsoftware.ch (import)
ge.submissions0
ge.oai.exportid148934
ge.oai.repositoryid58
ge.oai.setnameComputer Science
ge.oai.setspeccs
ge.oai.streamid2
ge.setnameGlobeEthicsLib
ge.setspecglobeethicslib
ge.linkhttp://arxiv.org/abs/1603.04350


This item appears in the following Collection(s)

Show simple item record