DC Field | Value | Language |
---|---|---|
dc.contributor.author | 김종민 | - |
dc.contributor.author | 김성웅 | - |
dc.contributor.author | 유창동 | - |
dc.date.accessioned | 2013-03-28T03:39:47Z | - |
dc.date.available | 2013-03-28T03:39:47Z | - |
dc.date.created | 2012-07-05 | - |
dc.date.issued | 2009-07 | - |
dc.identifier.citation | 대한전자공학회 하계종합학술대회 , v., no., pp.927 - 928 | - |
dc.identifier.uri | http://hdl.handle.net/10203/162624 | - |
dc.description.abstract | In this paper, we propose ν-structured support vector machine(ν -SSVM), an extension of -support vector machine(ν -SVM) to structured prediction problems. The proposed ν-SSVM is modified from the structured SVM(SSVM) to have both advantages of the intuitive parameter ν in the ν-SVM and the margin scaling in the SSVM. In the ν-SSVM, ν asymptotically equals the empirical risk over support vectors. The stochastic subgradient descent is used to solve the optimization problem of the ν-SSVM. | - |
dc.language | KOR | - |
dc.publisher | 대한전자공학회 | - |
dc.title | ν-Structured 서포트 벡터 머신을 이용한 음성 인식기 훈련 | - |
dc.title.alternative | ν-Structured Support Vector Machines For Phonetic Recognition | - |
dc.type | Conference | - |
dc.type.rims | CONF | - |
dc.citation.beginningpage | 927 | - |
dc.citation.endingpage | 928 | - |
dc.citation.publicationname | 대한전자공학회 하계종합학술대회 | - |
dc.identifier.conferencecountry | South Korea | - |
dc.contributor.localauthor | 유창동 | - |
dc.contributor.nonIdAuthor | 김종민 | - |
dc.contributor.nonIdAuthor | 김성웅 | - |
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.