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

Intelligent Sensor Data Analysis based on Cooperation of Knowledge Bases and Statistical Machine Learning

Research Project

Project/Area Number 19K21550
Project/Area Number (Other) 18H06487 (2018)
Research Category

Grant-in-Aid for Research Activity Start-up

Allocation TypeMulti-year Fund (2019)
Single-year Grants (2018)
Review Section 1002:Human informatics, applied informatics and related fields
Research InstitutionInstitute of Physical and Chemical Research

Principal Investigator

Takeishi Naoya  国立研究開発法人理化学研究所, 革新知能統合研究センター, 特別研究員 (20824030)

Project Period (FY) 2018-08-24 – 2021-03-31
Project Status Completed (Fiscal Year 2020)
Budget Amount *help
¥2,990,000 (Direct Cost: ¥2,300,000、Indirect Cost: ¥690,000)
Fiscal Year 2019: ¥1,430,000 (Direct Cost: ¥1,100,000、Indirect Cost: ¥330,000)
Fiscal Year 2018: ¥1,560,000 (Direct Cost: ¥1,200,000、Indirect Cost: ¥360,000)
Keywords統計的機械学習 / 知識ベース / 事前知識 / 専門家知識 / センサデータ解析 / ナレッジグラフ / 異常検知 / 統計的関係学習
Outline of Research at the Start

自動車,工場,人工衛星などの工学システムから取得することのできる多種多様なセンサデータは故障検知や運用最適化等に有用である.大量のセンサデータを有効活用するために統計的機械学習に基づく手法が注目されているが,機械学習の結果と専門家知識の整合性が判断しづらい等の理由でシステム運用上の意思決定には必ずしも活用されていない.そこで本研究では,工学システムの設計および運用上の知識(機器同士の関係性や運用上のルールなど)が知識ベースとして集約されていることに着目し,そのような知識ベースを統計的機械学習の事前知識として活用する方法を研究する.

Outline of Final Research Achievements

Statistial machine learning is a procedure to acquire (semi-)automatically a system that solves particular tasks with data of such tasks as input. Machine learning has been utilized in various tasks and data, but it is still challenging to interpret the results and to adapt to a small-data regime. We studied methodologies to incorpolate prior knowledge available as knowledge bases of application domains efficiently into machine learning. In particular, we focused on sensor data that often appear in engineering. We developed methods for incorporating prior knowledge, such as system diagrams (i.e., relation between feature of sensor data) and stability property of a system, into machine learning.

Academic Significance and Societal Importance of the Research Achievements

本研究では、センサデータ活用の場面で想定される形式の事前知識(システム図やシステムの安定性に関する知識)を機械学習に組み込む汎用的な方法を開発した。つまり、これまで利用することが難しかった、または利用するためには煩雑でアドホックな操作が必要だった事前知識を容易に機械学習で用いることができる。これにより、機械学習結果の効率や解釈性の向上が期待され、システム運用の場面で機械学習をさらに活用する助けになると期待できる。

Report

(4 results)
  • 2020 Annual Research Report   Final Research Report ( PDF )
  • 2019 Research-status Report
  • 2018 Annual Research Report
  • Research Products

    (11 results)

All 2021 2020 2019 2018

All Journal Article (6 results) (of which Peer Reviewed: 6 results,  Open Access: 4 results) Presentation (5 results)

  • [Journal Article] Learning Dynamics Models with Stable Invariant Sets2021

    • Author(s)
      Naoya Takeishi and Kawahara Yoshinobu
    • Journal Title

      Proceedings of the Thirty-Fifth AAAI Conference on Artificial Intelligence

      Volume: -

    • Related Report
      2020 Annual Research Report
    • Peer Reviewed / Open Access
  • [Journal Article] Knowledge-Based Regularization in Generative Modeling2020

    • Author(s)
      Naoya Takeishi、Yoshinobu Kawahara
    • Journal Title

      Proceedings of the 29th International Joint Conference on Artificial Intelligence

      Volume: -

    • Related Report
      2019 Research-status Report
    • Peer Reviewed / Open Access
  • [Journal Article] Kernel Learning for Data-Driven Spectral Analysis of Koopman Operators2019

    • Author(s)
      Naoya Takeishi
    • Journal Title

      Proceedings of Machine Learning Research

      Volume: 101 Pages: 956-971

    • Related Report
      2019 Research-status Report
    • Peer Reviewed / Open Access
  • [Journal Article] Shapley Values of Reconstruction Errors of PCA for Explaining Anomaly Detection2019

    • Author(s)
      Naoya Takeishi
    • Journal Title

      Proceedings of the 2019 International Conference on Data Mining Workshops

      Volume: - Pages: 793-798

    • DOI

      10.1109/icdmw.2019.00117

    • Related Report
      2019 Research-status Report
    • Peer Reviewed
  • [Journal Article] Neural Gray-Box Identification of Nonlinear Partial Differential Equations2019

    • Author(s)
      Sasaki Riku、Takeishi Naoya、Yairi Takehisa、Hori Koichi
    • Journal Title

      Lecture Notes in Computer Science

      Volume: 11671 Pages: 309-321

    • DOI

      10.1007/978-3-030-29911-8_24

    • ISBN
      9783030299101, 9783030299118
    • Related Report
      2018 Annual Research Report
    • Peer Reviewed
  • [Journal Article] Factorially Switching Dynamic Mode Decomposition for Koopman Analysis of Time-Variant Systems2018

    • Author(s)
      N. Takeishi, T. Yairi and Y. Kawahara
    • Journal Title

      Proceedings of 2018 IEEE Conference on Decision and Control (CDC'18)

      Volume: -- Pages: 6402-6408

    • DOI

      10.1109/cdc.2018.8619846

    • Related Report
      2018 Annual Research Report
    • Peer Reviewed / Open Access
  • [Presentation] 安定不変集合をもつ力学系の学習2020

    • Author(s)
      武石 直也, 河原 吉伸
    • Organizer
      第23回情報論的学習理論ワークショップ
    • Related Report
      2020 Annual Research Report
  • [Presentation] 再構成誤差のシャープレイ値による異常検知の説明2019

    • Author(s)
      武石 直也
    • Organizer
      第22回情報論的学習理論ワークショップ
    • Related Report
      2019 Research-status Report
  • [Presentation] 時変動的モード分解2019

    • Author(s)
      武石 直也
    • Organizer
      第33回人工知能学会全国大会
    • Related Report
      2019 Research-status Report
  • [Presentation] Knowledge-Based Distant Regularization in Learning Probabilistic Models2018

    • Author(s)
      Naoya Takeishi、Kosuke Akimoto
    • Organizer
      The 8th International Workshop on Statistical Relational AI
    • Related Report
      2018 Annual Research Report
  • [Presentation] 知識グラフによる生成モデル学習の正則化2018

    • Author(s)
      武石 直也
    • Organizer
      第21回情報論的学習理論ワークショップ
    • Related Report
      2018 Annual Research Report

URL: 

Published: 2018-08-27   Modified: 2024-03-26  

Information User Guide FAQ News Terms of Use Attribution of KAKENHI

Powered by NII kakenhi