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

Time-series Anomaly Detection based on Difference Subspace between Signal Subspaces

Research Project

Project/Area Number 19H04129
Research Category

Grant-in-Aid for Scientific Research (B)

Allocation TypeSingle-year Grants
Section一般
Review Section Basic Section 61010:Perceptual information processing-related
Research InstitutionUniversity of Tsukuba

Principal Investigator

Fukui Kazuhiro  筑波大学, システム情報系, 教授 (40375423)

Co-Investigator(Kenkyū-buntansha) 小林 匠  国立研究開発法人産業技術総合研究所, 情報・人間工学領域, 主任研究員 (30443188)
飯塚 里志  筑波大学, システム情報系, 准教授 (30755153)
Project Period (FY) 2019-04-01 – 2022-03-31
Project Status Completed (Fiscal Year 2022)
Budget Amount *help
¥17,290,000 (Direct Cost: ¥13,300,000、Indirect Cost: ¥3,990,000)
Fiscal Year 2021: ¥4,680,000 (Direct Cost: ¥3,600,000、Indirect Cost: ¥1,080,000)
Fiscal Year 2020: ¥5,070,000 (Direct Cost: ¥3,900,000、Indirect Cost: ¥1,170,000)
Fiscal Year 2019: ¥7,540,000 (Direct Cost: ¥5,800,000、Indirect Cost: ¥1,740,000)
Keywords時系列解析 / 変化検知 / 特異スペクトル解析 / 部分空間表現 / 差分部分空間 / 異常検知 / 時系列データ / 変化検出 / 部分空間 / 正準角 / 時系列 / 動画像列 / 凸錐差分部分空間 / 変化・異常検知 / 凸錐表現
Outline of Research at the Start

本研究では、時系列データから僅かな変化を検知し、その異常タイプを判別する新たな理論基盤を構築する。提案基盤の独自性は、時系列データの各時刻の状態を部分空間でコンパクトにモデル化し、その部分空間の幾何学的な構造変動を精密に捉えることにある。この枠組みをさらに強固にするために、時系列データは非負値を取る場合が多いことに着目して、非負制約を有する凸錐モデルと凸錐差分部分空間を導入する。提案基盤の有効性を、生体信号から動画像列まで様々なタイプの時系列データを用いて検証し、実応用展開に向けた理論基盤を確立する。

Outline of Final Research Achievements

We proposed a new method for anomaly detection in time-series data by incorporating the concept of difference subspace into the singular spectrum analysis (SSA). The key idea is to monitor slight temporal variations of the difference subspace between two signal subspaces corresponding to the past and present time-series data, as anomaly score. It is a natural generalization of the conventional SSA-based method which measures the minimum angle between the two signal subspaces as the degree of changes. By replacing the minimum angle with the difference subspace, our method boosted the performance while using the SSA-based framework as it can capture the whole structural difference between the two subspaces in its magnitude and direction. We demonstrated our method's effectiveness through performance evaluations on public time-series datasets.

Academic Significance and Societal Importance of the Research Achievements

近年,工場の生産ラインや社会インフラなどの複雑システムには,多種多様なセンサ群が配置されており,システム内部状態を反映した膨大な時系列データを得ることが可能となっている.しかしながら,データ量の増大と供にオペレーターの作業負担が増しており,これを軽減することは社会的なニーズが高い.本研究で取り組んだ時系列からの変化・異常検知は,データから通常と異なる僅かな時間変動を異常として自動検知することを可能とする.これによりオペレータの作業負担を大きく減らすと期待できる.

Report

(4 results)
  • 2022 Final Research Report ( PDF )
  • 2021 Annual Research Report
  • 2020 Annual Research Report
  • 2019 Annual Research Report
  • Research Products

    (19 results)

All 2023 2022 2021 2020 2019 Other

All Int'l Joint Research (2 results) Journal Article (10 results) (of which Int'l Joint Research: 3 results,  Peer Reviewed: 6 results,  Open Access: 2 results) Presentation (7 results) (of which Int'l Joint Research: 3 results)

  • [Int'l Joint Research] スウェーデン王立工科大学(スウェーデン)

    • Related Report
      2021 Annual Research Report
  • [Int'l Joint Research] ユニヴァーシティ・カレッジ・ロンドン(UCL)(英国)

    • Related Report
      2019 Annual Research Report
  • [Journal Article] Time-series Anomaly Detection based on Difference Subspace between Signal Subspaces2023

    • Author(s)
      Takumi Kanai, Naoya Sogi, Atsuto Maki, Kazuhiro Fukui
    • Journal Title

      ArXiv CoRR abs/2303.17802

      Volume: -

    • Related Report
      2021 Annual Research Report
    • Int'l Joint Research
  • [Journal Article] Temporal-stochastic tensor features for action recognition2022

    • Author(s)
      Batalo Bojan、Souza Lincon S.、Gatto Bernardo B.、Sogi Naoya、Fukui Kazuhiro
    • Journal Title

      Machine Learning with Applications

      Volume: 10 Pages: 100407-100407

    • DOI

      10.1016/j.mlwa.2022.100407

    • Related Report
      2021 Annual Research Report
    • Peer Reviewed / Open Access / Int'l Joint Research
  • [Journal Article] Analysis of Temporal Tensor Datasets on Product Grassmann Manifold2022

    • Author(s)
      Bojan Batalo, Lincon S. Souza, Bernardo B. Gatto, Naoya Sogi, Kazuhiro Fukui
    • Journal Title

      CVPR Workshops

      Volume: - Pages: 4868-4876

    • DOI

      10.1109/cvprw56347.2022.00534

    • Related Report
      2021 Annual Research Report
    • Peer Reviewed / Open Access
  • [Journal Article] 差分部分空間の幾何変動に基づく変化検知2022

    • Author(s)
      金井拓海, 枌 尚弥, 牧 淳人, 福井和広
    • Journal Title

      電子情報通信学会PRMU研究会

      Volume: 25 Pages: 18-23

    • Related Report
      2021 Annual Research Report
  • [Journal Article] Slow Feature Subspace for Action Recognition2021

    • Author(s)
      Suzana R. A. Beleza, Kazuhiro Fukui
    • Journal Title

      ICPR2020 Workshops proceedings

      Volume: 3 Pages: 702-716

    • Related Report
      2020 Annual Research Report
    • Peer Reviewed
  • [Journal Article] Fingerspelling Recognition with Two-Steps Cascade Process of Spotting and Classification2021

    • Author(s)
      Masanori Muroi, Naoya Sogi, Nobuko Kato, Kazuhiro Fukui
    • Journal Title

      ICPR2020 Workshops proceedings

      Volume: 6 Pages: 728-743

    • Related Report
      2020 Annual Research Report
    • Peer Reviewed
  • [Journal Article] 部分空間を用いたグラフ構造の表現学習法の提案2020

    • Author(s)
      石河純輝, 塩川浩昭, 福井和広
    • Journal Title

      信学会 情報論的学習理論と機械学習研究会(IBISML)予稿集

      Volume: 119 Pages: 51-57

    • Related Report
      2019 Annual Research Report
  • [Journal Article] 部分空間の幾何学構造の時間変動に基づく変化検知の提案2020

    • Author(s)
      金盛ほなみ, 枌 尚弥, 福井和広
    • Journal Title

      信学会 情報論的学習理論と機械学習研究会(IBISML)予稿集

      Volume: 119 Pages: 83-89

    • Related Report
      2019 Annual Research Report
  • [Journal Article] Enhanced Grassmann discriminant analysis with randomized time warping for motion recognition2020

    • Author(s)
      Souza, Lincon Sales de Gatto, Bernardo B. Xue, Jing-Hao Fukui, Kazuhiro
    • Journal Title

      Pattern recognition

      Volume: 97 Pages: 1-11

    • DOI

      10.1016/j.patcog.2019.107028

    • Related Report
      2019 Annual Research Report
    • Peer Reviewed / Int'l Joint Research
  • [Journal Article] 凸錐判別分析に基づく画像セットベース識別2019

    • Author(s)
      枌尚弥, Lincon Souza, Bernardo Gatto, Jing-Hao Xue, Rui Zhu, 福井和広
    • Journal Title

      画像の認識・理解シンポジウム(MIRU2019) 予稿集

      Volume: -

    • Related Report
      2019 Annual Research Report
    • Peer Reviewed
  • [Presentation] Analysis of Temporal Tensor Datasets on Product Grassmann Manifold2022

    • Author(s)
      Bojan Batalo
    • Organizer
      CVPR Workshops
    • Related Report
      2021 Annual Research Report
    • Int'l Joint Research
  • [Presentation] 差分部分空間の幾何変動に基づく変化検知2022

    • Author(s)
      金井拓海
    • Organizer
      電子情報通信学会PRMU研究会
    • Related Report
      2021 Annual Research Report
  • [Presentation] 差分部分空間を用いた特異スペクトル解析に基づく変化検知2022

    • Author(s)
      金井拓海
    • Organizer
      第25回情報論的学習理論ワークショップ
    • Related Report
      2021 Annual Research Report
  • [Presentation] Slow Feature Subspace for Action Recognition2021

    • Author(s)
      Suzana R. A. Beleza
    • Organizer
      ICPR2020 Workshops
    • Related Report
      2020 Annual Research Report
    • Int'l Joint Research
  • [Presentation] Fingerspelling Recognition with Two-Steps Cascade Process of Spotting and Classification2021

    • Author(s)
      Masanori Muroi
    • Organizer
      ICPR2020 Workshops
    • Related Report
      2020 Annual Research Report
    • Int'l Joint Research
  • [Presentation] 部分空間の幾何学構造の時間変動に基づく変化検知の提案2020

    • Author(s)
      金盛ほなみ
    • Organizer
      信学会 情報論的学習理論と機械学習研究会(IBISML)
    • Related Report
      2019 Annual Research Report
  • [Presentation] 凸錐判別分析に基づく画像セットベース識別2019

    • Author(s)
      枌尚弥
    • Organizer
      画像の認識・理解シンポジウム(MIRU2019)
    • Related Report
      2019 Annual Research Report

URL: 

Published: 2019-04-18   Modified: 2024-01-30  

Information User Guide FAQ News Terms of Use Attribution of KAKENHI

Powered by NII kakenhi