Leveraging imperfect grouping and partial support sharing in multitask learning태스크의 불완전한 그룹화와 서포트 부분 공유를 이용한 멀티 태스크 학습

Cited 0 time in webofscience Cited 0 time in scopus
  • Hit : 539
  • Download : 0
DC FieldValueLanguage
dc.contributor.advisorYang, Eunho-
dc.contributor.advisor양은호-
dc.contributor.authorHong, Youngki-
dc.date.accessioned2019-09-04T02:47:01Z-
dc.date.available2019-09-04T02:47:01Z-
dc.date.issued2019-
dc.identifier.urihttp://library.kaist.ac.kr/search/detail/view.do?bibCtrlNo=843549&flag=dissertationen_US
dc.identifier.urihttp://hdl.handle.net/10203/267059-
dc.description학위논문(석사) - 한국과학기술원 : 전산학부, 2019.2,[iv, 26 p. :]-
dc.description.abstractWe consider the problem of simultaneously learning multiple sparse representations in the high-dimensional setting, exemplified by Multitask Learning (MTL). We propose a new estimator, which we call Dirty Fusion (DF). DF bridges the gap between dirty models, which decompose parameters into common and model-specific sets, and grouping approaches, which assume models belonging to the same group share the same sparsity pattern or have similar parameter values. DF jointly estimates the model parameters together with their potentially “unclean” group structures, and allows for partial support overlap within each group. We formulate DF estimators as an optimization problem, and incorporate automatic debiasing variables into the learning formulation. We demonstrate the impact of the approach using synthetic and real data.-
dc.languageeng-
dc.publisher한국과학기술원-
dc.subjectmultitask learning▼amultitask group learning▼afusion regularization▼adirty Model-
dc.subject멀티 태스크 학습▼a멀티 태스크 그룹 학습▼a퓨전 정규화▼a합성 모델-
dc.titleLeveraging imperfect grouping and partial support sharing in multitask learning-
dc.title.alternative태스크의 불완전한 그룹화와 서포트 부분 공유를 이용한 멀티 태스크 학습-
dc.typeThesis(Master)-
dc.identifier.CNRN325007-
dc.description.department한국과학기술원 :전산학부,-
dc.contributor.alternativeauthor홍영기-
Appears in Collection
CS-Theses_Master(석사논문)
Files in This Item
There are no files associated with this item.

qr_code

  • mendeley

    citeulike


rss_1.0 rss_2.0 atom_1.0