Presentation 2017-11-16
Document classification on unknown class using transfer learning
Masato Shirai, Jianquan Liu,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) In this investigation, we propose a new classification method based on transfer learning to classify new classes without training data of that class. Generally,features of classes is obtained from labelled data. However, large amount of labeled data is required to obtain features of classes . Transfer learning utilizes extracted knowledge from source domain to analyze the target domain. We extract latent topics based on topic model from source domain. And we select appropriate source domain corresponding to class of target domain using features of class tag. We estimate a class of document in target domain by class information of source domain.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Transfer learning / Document classification / Topic model
Paper # CAS2017-48,MSS2017-32
Date of Issue 2017-11-09 (CAS, MSS)

Conference Information
Committee CAS / MSS / IPSJ-AL
Conference Date 2017/11/16(2days)
Place (in Japanese) (See Japanese page)
Place (in English)
Topics (in Japanese) (See Japanese page)
Topics (in English)
Chair Mitsuru Hiraki(Renesas) / Morikazu Nakamura(Univ. of Ryukyus) / Takashi Horiyama(Saitama University)
Vice Chair Hideaki Okazaki(Shonan Inst. of Tech.) / Shigemasa Takai(Osaka Univ.)
Secretary Hideaki Okazaki(Renesas) / Shigemasa Takai(Shonan Inst. of Tech.) / (Toshiba)
Assistant Yohei Nakamura(Hitachi) / Hideki Kinjo(Okinawa Univ.)

Paper Information
Registration To Technical Committee on Circuits and Systems / Technical Committee on Mathematical Systems Science and its applications / Special Interest Group on Algorithms
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Document classification on unknown class using transfer learning
Sub Title (in English)
Keyword(1) Transfer learning
Keyword(2) Document classification
Keyword(3) Topic model
1st Author's Name Masato Shirai
1st Author's Affiliation Shimane University(Shimane Univ.)
2nd Author's Name Jianquan Liu
2nd Author's Affiliation NEC Corporation(NEC)
Date 2017-11-16
Paper # CAS2017-48,MSS2017-32
Volume (vol) vol.117
Number (no) CAS-300,MSS-301
Page pp.pp.51-56(CAS), pp.51-56(MSS),
#Pages 6
Date of Issue 2017-11-09 (CAS, MSS)