Study on Information theoretic Learning Theory of Latent Dynamics

2015 Fiscal Year Final Research Report

• Project/Area Number: 23240019
• Research Category: Grant-in-Aid for Scientific Research (A)
• Allocation Type: Single-year Grants
• Section: 一般
• Research Field: Intelligent informatics
• Research Institution: The University of Tokyo
• Principal Investigator: Yamanishi Kenji 東京大学, 情報理工学(系)研究科, 教授 (90549180)
• Co-Investigator(Kenkyū-buntansha): KASHIMA Hisashi 京都大学, 情報学研究科, 教授 (80545583) ; TOMIOKA Ryota Microsoft Research, Researcher (70518282)
• Research Collaborator: TERAZONO Yasushi ; SAKURAI Ei-ichi ; URABE Yasuhiro ; IWAI Hiroki ; TAKAHASHI Toshimitsu ; HIRAI So ; HAYASHI Yu ; KANAZAWA Hiroki ; SATO Sho-ichi ; SAKAI Yoshiki ; LIANG Zenghan ; OEDA Shin-ichi ; ASAOKA Ryo ; MURATA Hiroshi ; SAITO Shota ; Ito Yu ; MIYAGUCHI Kohei ; KAJIMURA Shun-suke ; BABA Yukino ; OYAMA Satohi ; KURIHARA Masahito ; WANG Jingling
• Project Period (FY): 2011-04-01 – 2016-03-31
• Keywords: 情報論的学習理論 / データマイニング / 潜在的ダイナミクス / 機械学習 / ビッグデータ

Outline of Final Research Achievements

Data mining technologies for extracting valuable information from big data are significantly important nowadays. The main purpose of conventional data mining technologies is to extract superficial statistical patterns from data. We are rather concerned with the issue of how to learn latent information behind data and how to detect its changes, which we call “latent dynamics.” We construct a theory for learning latent dynamics from a unifying view of information-theoretic learning theory, specifically on the basis of the minimum description length principle. We also demonstrate its effectiveness through the applications to real world data (e.g. security, SNS, marketing, healthcare, education, etc.).

Free Research Field

データマイニング