• Search Research Projects
  • Search Researchers
  • How to Use
  1. Back to previous page

Variable Selection for Small Sample and High Dimension Case by Semi-supervised Learning and Its Application to Super-Resolution

Research Project

Project/Area Number 25870503
Research Category

Grant-in-Aid for Young Scientists (B)

Allocation TypeMulti-year Fund
Research Field Statistical science
Perceptual information processing
Research InstitutionKyushu University

Principal Investigator

Kawakita Masanori  九州大学, システム情報科学研究院, 助教 (90435496)

Project Period (FY) 2013-04-01 – 2017-03-31
Project Status Completed (Fiscal Year 2016)
Budget Amount *help
¥4,290,000 (Direct Cost: ¥3,300,000、Indirect Cost: ¥990,000)
Fiscal Year 2015: ¥1,430,000 (Direct Cost: ¥1,100,000、Indirect Cost: ¥330,000)
Fiscal Year 2014: ¥1,560,000 (Direct Cost: ¥1,200,000、Indirect Cost: ¥360,000)
Fiscal Year 2013: ¥1,300,000 (Direct Cost: ¥1,000,000、Indirect Cost: ¥300,000)
Keywords半教師付き学習 / MDL原理 / 超解像 / Barron and Cover理論 / スパースコーディング / 最小記述長原理 / 共変量シフト / リスクバウンド / 重み付き尤度法 / lasso / リスク上界 / MDL / 影響関数 / 最有効推定量 / セミパラメトリック理論
Outline of Final Research Achievements

The main important result of this study is that we provided a way to extend Barron and Cover’s theory to supervised learning without any significant lack of its virtues, which had been considered to be difficult. Our extension leads to a risk estimator of supervised learning without conventional assumptions like boundedness of random variables and/or asymptotic assumption. By our method, we succeeded in deriving a new risk bound of the most famous compressed sensing algorithm (lasso). We also extended these results to semi-supervised learning and sparse coding. Furthermore, by implementing semi-supervised sparse coding, we construct a new semi-supervised super-resolution algorithm. We show that the accuracy of super-resolution can be improved by semi-supervised super-resolution by numerical experiments though its extent strongly depends on input images.

Report

(5 results)
  • 2016 Annual Research Report   Final Research Report ( PDF )
  • 2015 Research-status Report
  • 2014 Research-status Report
  • 2013 Research-status Report
  • Research Products

    (13 results)

All 2017 2016 2015 2014

All Journal Article (1 results) (of which Peer Reviewed: 1 results) Presentation (12 results) (of which Int'l Joint Research: 2 results,  Invited: 2 results)

  • [Journal Article] Safe semi-supervised learning based on weighted likelihood2014

    • Author(s)
      Kawakita, M. and Takeuchi, J.
    • Journal Title

      Neural Networks

      Volume: 53 Pages: 146-164

    • Related Report
      2013 Research-status Report
    • Peer Reviewed
  • [Presentation] スパースコーディングの転移学習と超解像への応用2017

    • Author(s)
      川喜田雅則
    • Organizer
      JST CRESTシンポジウム「ビッグデータ利活用のための基盤構築とその応用」
    • Place of Presentation
      名古屋工業大学
    • Year and Date
      2017-02-16
    • Related Report
      2016 Annual Research Report
    • Invited
  • [Presentation] Barron and Cover’s Theory in Supervised Learning and Its Application to Lasso2016

    • Author(s)
      M. Kawakita and J. Takeuchi
    • Organizer
      International Conference on Machine Learning 2016
    • Place of Presentation
      New York, U.S.A.
    • Year and Date
      2016-06-20
    • Related Report
      2016 Annual Research Report
    • Int'l Joint Research
  • [Presentation] Barron and Covers’ Theory in Supervised Learning and Its Application to Lasso2016

    • Author(s)
      Masanori Kawakita and Jun'ichi Takeuchi
    • Organizer
      International Conference on Machine Learning 2016
    • Place of Presentation
      New York, USA
    • Year and Date
      2016-06-19
    • Related Report
      2015 Research-status Report
    • Int'l Joint Research
  • [Presentation] Barron and Cover 理論によるlasso のリスクの上界評価2016

    • Author(s)
      川喜田 雅則
    • Organizer
      2016年IEICE総合大会チュートリアルセッション「記述長最小原理の新展開」
    • Place of Presentation
      九州大学伊都キャンパス
    • Year and Date
      2016-03-15
    • Related Report
      2015 Research-status Report
  • [Presentation] 教師付き学習におけるMDL原理とlassoのリスク評価2016

    • Author(s)
      川喜田雅則, 竹内純一
    • Organizer
      科研費シンポジウム「統計学と機械学習における数理とモデリング」
    • Place of Presentation
      東京工業大学大岡山キャンパス
    • Year and Date
      2016-02-21
    • Related Report
      2015 Research-status Report
    • Invited
  • [Presentation] 教師付き学習におけるMDL推定のリスクバウンドについて2015

    • Author(s)
      川喜田雅則, 竹内純一
    • Organizer
      第37回情報理論とその応用シンポジウム
    • Place of Presentation
      岡山県児島
    • Year and Date
      2015-11-24
    • Related Report
      2015 Research-status Report
  • [Presentation] 教師付き学習におけるMDL推定2015

    • Author(s)
      竹内純一, 川喜田雅則
    • Organizer
      第9回シャノン理論ワークショップ
    • Place of Presentation
      渡瀬温泉心の宿 わたらせ温泉
    • Year and Date
      2015-09-25
    • Related Report
      2015 Research-status Report
  • [Presentation] 教師付き学習におけるMDL推定量の設計とその収束速度2015

    • Author(s)
      竹内純一, 川喜田雅則
    • Organizer
      電子情報通信学会技術報告 IT2015-26
    • Place of Presentation
      東京工業大学
    • Year and Date
      2015-07-13
    • Related Report
      2015 Research-status Report
  • [Presentation] MDL理論によるlassoのリスク上界2015

    • Author(s)
      川喜田雅則,豊暉原侑心,竹内純一
    • Organizer
      第21回IBISML研究会
    • Place of Presentation
      沖縄科学技術大学院大学
    • Year and Date
      2015-06-23
    • Related Report
      2015 Research-status Report
  • [Presentation] L1罰則付き線形回帰のMDLによる推定誤差上界について2015

    • Author(s)
      豊暉原侑心,川喜田雅則,竹内純一
    • Organizer
      IT研究会
    • Place of Presentation
      北九州市立大学
    • Year and Date
      2015-03-03
    • Related Report
      2014 Research-status Report
  • [Presentation] 半教師付き学習の影響関数からの推定方程式の復元について2014

    • Author(s)
      川喜田雅則, 藤澤洋徳
    • Organizer
      第17回情報論的学習理論ワークショップ
    • Place of Presentation
      名古屋大学
    • Year and Date
      2014-11-18
    • Related Report
      2014 Research-status Report
  • [Presentation] 半教師付き学習における影響関数から推定方程式の構築について2014

    • Author(s)
      川喜田雅則, 藤澤洋徳
    • Organizer
      2014年度統計関連学会連合大会
    • Place of Presentation
      東京大学
    • Year and Date
      2014-09-14
    • Related Report
      2014 Research-status Report

URL: 

Published: 2014-07-25   Modified: 2019-07-29  

Information User Guide FAQ News Terms of Use Attribution of KAKENHI

Powered by NII kakenhi