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

探索的分析によるデータ駆動型仮説の信頼性評価法の確立と生命科学分野における実証

Research Project

Project/Area Number 20H00601
Research Category

Grant-in-Aid for Scientific Research (A)

Allocation TypeSingle-year Grants
Section一般
Review Section Medium-sized Section 61:Human informatics and related fields
Research InstitutionNagoya University (2022-2024)
Nagoya Institute of Technology (2020-2021)

Principal Investigator

竹内 一郎  名古屋大学, 工学研究科, 教授 (40335146)

Co-Investigator(Kenkyū-buntansha) 花田 博幸  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, 研究員 (00793035)
寺田 吉壱  大阪大学, 大学院基礎工学研究科, 准教授 (10738793)
稲津 佑  名古屋工業大学, 工学(系)研究科(研究院), 助教 (20869896)
本谷 秀堅  名古屋工業大学, 工学(系)研究科(研究院), 教授 (60282688)
津田 宏治  東京大学, 大学院新領域創成科学研究科, 教授 (90357517)
Project Period (FY) 2020-04-01 – 2025-03-31
Project Status Granted (Fiscal Year 2024)
Budget Amount *help
¥43,290,000 (Direct Cost: ¥33,300,000、Indirect Cost: ¥9,990,000)
Fiscal Year 2024: ¥6,110,000 (Direct Cost: ¥4,700,000、Indirect Cost: ¥1,410,000)
Fiscal Year 2023: ¥10,010,000 (Direct Cost: ¥7,700,000、Indirect Cost: ¥2,310,000)
Fiscal Year 2022: ¥10,010,000 (Direct Cost: ¥7,700,000、Indirect Cost: ¥2,310,000)
Fiscal Year 2021: ¥10,920,000 (Direct Cost: ¥8,400,000、Indirect Cost: ¥2,520,000)
Fiscal Year 2020: ¥6,240,000 (Direct Cost: ¥4,800,000、Indirect Cost: ¥1,440,000)
Keywords機械学習 / データサイエンス / 統計的仮説検定 / 統計科学 / 人工知能 / 選択的推論 / 生命科学 / 教師なし学習 / 統計的推測 / 医療情報学
Outline of Research at the Start

機械学習などのデータ科学に基づく科学研究はデータ駆動型科学と呼ばれ,データ分析によって仮説を自動生成するため,研究者の知識や経験からは想起できないような新しい仮説を生み出せる可能性がある.しかし,複雑なデータを複雑なアルゴリズムで分析して得られる「データ駆動型仮説」はデータに過剰に適合している可能性があり,信頼性評価が難しい.生命医療分野など誤った判断のもたらすリスクが大きい場合,データ駆動型仮説の信頼性を適切に評価する枠組が不可欠である.本研究では,探索的データ分析によって得られたデータ駆動型仮説の信頼性評価を行う方法を開発する.

Outline of Annual Research Achievements

機械学習などのデータ分析技術を利用する科学研究のアプローチはデータ駆動型科学と呼ばれさまざまな分野で有望視されている.データ駆動型科学では研究対象に関するデータを分析することによって科学的仮説を生成するため,従来のアプローチでは思いつかないような仮説を生み出せる可能性がある. 一方,複雑なデータを複雑なアルゴリズムで分析して得られた仮説の信頼性を評価するのは難しい.特に,教師なし学習と呼ばれる探索的なデータ分析によって仮説が生成される場合,信頼性を保証する方法は確立されていない.特に,生命医療分野など,誤った判断のもたらすリスクが大きい状況ではデータ駆動型仮説の信頼性を確保することが不可欠である.本研究では,探索的データ分析によって得られたデータ駆動型仮説の信頼性を定量化する数理・情報基盤を構築し,その有用性を生命医療分野において実証する.教師なし学習アルゴリズムが強力であればあるほど(データへの適合力が大きければ大きいほど),有望な仮説が生成できる可能性が高まる一方,仮説選択バイアスも大きくなってしまう.データ駆動型仮説の信頼性保証は,仮説選択バイアスを正しく定量化し,その補正を行うことによって実現できる.本研究では仮説選択バイアスを適切に補正し,アルゴリズムが仮説を生成したという条件のもとで統計的推論(仮説検定の枠組による偽陽性率(p値)や信頼区間の計算)を行う方法を確立する.2023年度は,深層生成モデルによって生成された仮説に対する信頼性保証を行うための選択的推論の理論構築,アルゴリズム開発にとりくんだ.特に,開発した方法を異常検知の問題へ適用した.生成モデルを用いた異常検知では,異常を含む画像から正常画像を生成し,その差分によって異常を同定することができるが,開発した方法を用いると,この異常検知の結果に対して理論的妥当性を持つp値を計算することができる.

Current Status of Research Progress
Current Status of Research Progress

2: Research has progressed on the whole more than it was originally planned.

Reason

2023年度は,方法面では,前年度に引き続き,深層学習によって駆動される知識の信頼性評価法を選択的推論の枠組で発展させた.これまでは,予測のための深層学習モデルを対象としてきたが,今年度の研究では,生成のための深層学習モデルを対象とした.特に,開発した方法を異常検知の問題へ適用した.生成モデルを用いた異常検知では,異常を含む画像から正常画像を生成し,その差分によって異常を同定することができるが,開発した方法を用いると,この異常検知の結果に対して理論的妥当性を持つp値を計算することができる利点を持つ.本研究に関連する成果は,機械学習や関連分野の難関国際会議であるICMLなどに採択され,国内外から注目を集めている.

Strategy for Future Research Activity

2024年度以降も,選択的推論を基軸とし,方法面の発展と適用先の開拓を並列して実施する予定である.特に、Diffusion Modelなどの深層生成モデルに対する選択的推論を行えるように適用範囲を拡げること、また,実装コストを減らすためのライブラリ開発にに取り組む。また、新たな問題設定として、ドメイン適応後のデータ分析に対する選択的推論の理論構築,アルゴリズム開発,ソフトウェア設計を行う.

Report

(5 results)
  • 2023 Annual Research Report
  • 2022 Annual Research Report
  • 2021 Annual Research Report
  • 2020 Comments on the Screening Results   Annual Research Report
  • Research Products

    (37 results)

All 2024 2023 2022 2021 2020

All Journal Article (23 results) (of which Int'l Joint Research: 5 results,  Peer Reviewed: 23 results,  Open Access: 17 results) Presentation (14 results) (of which Int'l Joint Research: 13 results,  Invited: 1 results)

  • [Journal Article] A Confidence Machine for Sparse High-Order Interaction Model2024

    • Author(s)
      Das D., Ndiaye E. and Takeuchi I.
    • Journal Title

      Stat

      Volume: -

    • Related Report
      2023 Annual Research Report
    • Peer Reviewed / Open Access / Int'l Joint Research
  • [Journal Article] Case-based Similar Image Retrieval for Weakly Annotated Large Histopathological Images of Malignant Lymphoma Using Deep Metric Learning2023

    • Author(s)
      Hashimoto N., Takagi Y., Masuda H., Miyoshi H., Kohno K., Nagaishi M., Sato K., Takeuchi M., Furuta T., Kawamoto K., Yamada K., Moritsubo M., Inoue K., Shimasaki Y., Ogura Y., Imamoto T., Mishina T., Tanaka K., Kawaguchi Y., Nakamura S., Ohshima K., Hontani H., Takeuchi I.
    • Journal Title

      Medical Image Analysis

      Volume: -

    • Related Report
      2023 Annual Research Report
    • Peer Reviewed / Open Access / Int'l Joint Research
  • [Journal Article] Generalized Low-Rank Update: Model Parameter Bounds for Low-Rank Training Data Modifications2023

    • Author(s)
      Hanada H., Hashimoto N., Taji K. and Takeuchi I.
    • Journal Title

      Neural Computation

      Volume: -

    • Related Report
      2023 Annual Research Report
    • Peer Reviewed / Open Access / Int'l Joint Research
  • [Journal Article] Root-finding Approaches for Computing Conformal Prediction Set.2022

    • Author(s)
      Ndiaye E., Takeuchi I.
    • Journal Title

      Machine Learning

      Volume: -

    • Related Report
      2022 Annual Research Report
    • Peer Reviewed / Open Access / Int'l Joint Research
  • [Journal Article] More Powerful Conditional Selective Inference for Generalized Lasso by Parametric Programming.2022

    • Author(s)
      Duy V.N.L., Takeuchi I.
    • Journal Title

      Journal of Machine Learning Research

      Volume: -

    • Related Report
      2022 Annual Research Report
    • Peer Reviewed / Open Access
  • [Journal Article] Conditional Selective Inference for Robust Regression and Outlier Detection using Piecewise-Linear Homotopy Continuation.2022

    • Author(s)
      Tsukurimichi T., Inatsu Y., Duy V.N.L., Takeuchi I.
    • Journal Title

      Annals of Institute of Statistical Mathematics

      Volume: -

    • Related Report
      2022 Annual Research Report
    • Peer Reviewed / Open Access
  • [Journal Article] Subtype Classification of Malignant Lymphoma Using Immunohistochemical Staining Pattern2022

    • Author(s)
      Noriaki Hashimoto, Kaho Ko, Tatsuya Yokota, Kei Kohno, Masato Nakaguro, Shigeo Nakamura, Ichiro Takeuchi, Hidekata Hontani
    • Journal Title

      International Journal of Computer Assisted Radiology and Surgery

      Volume: 17

    • Related Report
      2021 Annual Research Report
    • Peer Reviewed / Open Access
  • [Journal Article] Fast and More Powerful Selective Inference for Sparse High-order Interaction Model2022

    • Author(s)
      Diptesh Das, Vo Nguyen Le Duy, Hiroyuki Hanada, Koji Tsuda, Ichiro Takeuchi
    • Journal Title

      Proceedings of AAAI Conference on Artificial Intelligence (AAAI2022)

      Volume: NA

    • Related Report
      2021 Annual Research Report
    • Peer Reviewed / Open Access
  • [Journal Article] Bayesian Quadrature Optimization for Probability Threshold Robustness Measure2021

    • Author(s)
      Shogo Iwazaki, Yu Inatsu, Ichiro Takeuchi
    • Journal Title

      Neural Computation

      Volume: 33 Pages: 3413-3466

    • Related Report
      2021 Annual Research Report
    • Peer Reviewed
  • [Journal Article] Distance Metric Learning for Graph Structured Data2021

    • Author(s)
      Tomoki Yoshida, Ichiro Takeuchi, Masayuki Karasuyama
    • Journal Title

      Machine Learning

      Volume: 110 Pages: 1765-1811

    • Related Report
      2021 Annual Research Report
    • Peer Reviewed
  • [Journal Article] Parametric Programming Approach for More Powerful and General Lasso Selective Inference2021

    • Author(s)
      Vo Nguyen Le Duy, Ichiro Takeuchi
    • Journal Title

      Proceedings of International Conference on Artifical Intelligence and Statistics (AISTATS2021)

      Volume: NA

    • Related Report
      2021 Annual Research Report
    • Peer Reviewed / Open Access
  • [Journal Article] Mean-Variance Analysis in Bayesian Optimization under Uncertainty2021

    • Author(s)
      Shogo Iwazaki, Yu Inatsu, Ichiro Takeuchi
    • Journal Title

      Proceedings of International Conference on Artifical Intelligence and Statistics (AISTATS2021)

      Volume: NA

    • Related Report
      2021 Annual Research Report
    • Peer Reviewed / Open Access
  • [Journal Article] Active Learning for Distributionally Robust Level-Set Estimation2021

    • Author(s)
      Yu Inatsu, Shogo Iwazaki, Ichiro Takeuchi
    • Journal Title

      Proceedings of International Conference on Machine Learning 2021 (ICML2021)

      Volume: NA

    • Related Report
      2021 Annual Research Report
    • Peer Reviewed / Open Access
  • [Journal Article] More Powerful and General Selective Inference for Stepwise Feature Selection using Homotopy Method2021

    • Author(s)
      Kazuya Sugiyama, Vo Nguyen Le Duy, Ichiro Takeuchi
    • Journal Title

      Proceedings of International Conference on Machine Learning 2021 (ICML2021)

      Volume: NA

    • Related Report
      2021 Annual Research Report
    • Peer Reviewed / Open Access
  • [Journal Article] Exploration of natural red-shifted rhodopsins using a machine learning-based Bayesian experimental design2021

    • Author(s)
      Inoue Keiichi、Karasuyama Masayuki、Nakamura Ryoko、Konno Masae、Yamada Daichi、Mannen Kentaro、Nagata Takashi、Inatsu Yu、Yawo Hiromu、Yura Kei、Beja Oded、Kandori Hideki、Takeuchi Ichiro
    • Journal Title

      Communications Biology

      Volume: 4 Issue: 1 Pages: 362-362

    • DOI

      10.1038/s42003-021-01878-9

    • Related Report
      2020 Annual Research Report
    • Peer Reviewed / Open Access / Int'l Joint Research
  • [Journal Article] Selective inference for high-order interaction features selected in a stepwise manner2021

    • Author(s)
      Shinya Suzumura, Kazuya Nakagawa, Yuta Umezu, Koji Tsuda, Ichiro Takeuchi
    • Journal Title

      IPSJ Transactions on Bioinformatics

      Volume: 14 Issue: 0 Pages: 1-11

    • DOI

      10.2197/ipsjtbio.14.1

    • NAID

      130007985966

    • Related Report
      2020 Annual Research Report
    • Peer Reviewed
  • [Journal Article] Bayesian Experimental Design for Finding Reliable Level Set under Input Uncertainty2020

    • Author(s)
      Shogo Iwazaki, Yu Inatsu, Ichiro Takeuchi.
    • Journal Title

      IEEE Access

      Volume: 8 Pages: 203982-203993

    • DOI

      10.1109/access.2020.3036863

    • Related Report
      2020 Annual Research Report
    • Peer Reviewed / Open Access
  • [Journal Article] Computing Valid p-value for Optimal Changepoint by Selective Inference using Dynamic Programming.2020

    • Author(s)
      Vo Nguyen Le Duy, Hiroki Toda, Ryota Sugiyama, Ichiro Takeuchi
    • Journal Title

      Proceedings of 34th Conference on Neural Information Processing Systems (NeurIPS2020)

      Volume: NA

    • Related Report
      2020 Annual Research Report
    • Peer Reviewed / Open Access
  • [Journal Article] Active Learning for Level Set Estimation Under Input Uncertainty and Its Extensions2020

    • Author(s)
      Yu Inatsu, Masayuki Karasuyama, Keiichi Inoue, Ichiro Takeuchi
    • Journal Title

      Neural Computation

      Volume: 32 Issue: 12 Pages: 2486-2531

    • DOI

      10.1162/neco_a_01332

    • Related Report
      2020 Annual Research Report
    • Peer Reviewed
  • [Journal Article] Active Learning of Bayesian Linear Models with High-Dimensional Binary Features by Parameter Confidence-Region Estimation2020

    • Author(s)
      Inatsu Yu、Karasuyama Masayuki、Inoue Keiichi、Kandori Hideki、Takeuchi Ichiro
    • Journal Title

      Neural Computation

      Volume: 32 Issue: 10 Pages: 1998-2031

    • DOI

      10.1162/neco_a_01310

    • Related Report
      2020 Annual Research Report
    • Peer Reviewed
  • [Journal Article] Active Learning for Enumerating Local Minima Based on Gaussian Process Derivatives2020

    • Author(s)
      Inatsu Yu、Sugita Daisuke、Toyoura Kazuaki、Takeuchi Ichiro
    • Journal Title

      Neural Computation

      Volume: 32 Issue: 10 Pages: 2032-2068

    • DOI

      10.1162/neco_a_01307

    • Related Report
      2020 Annual Research Report
    • Peer Reviewed
  • [Journal Article] Computing Valid P-Values for Image Segmentation by Selective Inference2020

    • Author(s)
      Tanizaki Kosuke、Hashimoto Noriaki、Inatsu Yu、Hontani Hidekata、Takeuchi Ichiro
    • Journal Title

      Proceedings of IEEE/CVF Conference on Computer Vision and Pattern Recognition 2020 (CVPR2020)

      Volume: - Pages: 9550-9559

    • DOI

      10.1109/cvpr42600.2020.00957

    • Related Report
      2020 Annual Research Report
    • Peer Reviewed / Open Access
  • [Journal Article] Multi-scale Domain-adversarial Multiple-instance CNN for Cancer Subtype Classification with Unannotated Histopathological Images2020

    • Author(s)
      Hashimoto Noriaki、Fukushima Daisuke、Koga Ryoichi、Takagi Yusuke、Ko Kaho、Kohno Kei、Nakaguro Masato、Nakamura Shigeo、Hontani Hidekata、Takeuchi Ichiro
    • Journal Title

      Proceedings of IEEE/CVF Conference on Computer Vision and Pattern Recognition 2020 (CVPR2020)

      Volume: - Pages: 3851-3860

    • DOI

      10.1109/cvpr42600.2020.00391

    • Related Report
      2020 Annual Research Report
    • Peer Reviewed / Open Access
  • [Presentation] Multi-objective Bayesian Optimization with Active Preference Learning2024

    • Author(s)
      Ozaki R., Ishikawa K., Kanzaki Y., Takeno S., Takeuchi I., Karasuyama M.
    • Organizer
      AAAI Conference on Artificial Intelligence
    • Related Report
      2023 Annual Research Report
    • Int'l Joint Research
  • [Presentation] A Stopping Criterion for Bayesian Optimization by The Gap of Expected Minimum Simple Regrets2023

    • Author(s)
      Ishibashi H., Karasuyama M., Takeuchi I., Hino H.
    • Organizer
      The International Conference on AI and Statistics
    • Related Report
      2023 Annual Research Report
    • Int'l Joint Research
  • [Presentation] Valid P-Value for Deep Learning-driven Salient Region2023

    • Author(s)
      Miwa D., Duy V.N.L., Takeuchi I.
    • Organizer
      International Conference on Learning Representation
    • Related Report
      2023 Annual Research Report
    • Int'l Joint Research
  • [Presentation] Quantifying Statistical Significance of Neural Network-based Image Segmentation by Selective Inference.2022

    • Author(s)
      Duy V.N.L., Iwazaki S., Takeuchi I.
    • Organizer
      Neural Information Processing Systems (NeurIPS)
    • Related Report
      2022 Annual Research Report
    • Int'l Joint Research
  • [Presentation] Bayesian Optimization for Distributionally Robust Chance-constrained Problem.2022

    • Author(s)
      Inatsu Y., Takeno S., Karasuyama M., Takeuchi I.
    • Organizer
      International Conference on Machine Learning (ICML)
    • Related Report
      2022 Annual Research Report
    • Int'l Joint Research
  • [Presentation] Fast and More Powerful Selective Inference for Sparse High-order Interaction Model2022

    • Author(s)
      Diptesh Das, Vo Nguyen Le Duy, Hiroyuki Hanada, Koji Tsuda, Ichiro Takeuchi
    • Organizer
      AAAI Conference on Artificial Intelligence (AAAI2022)
    • Related Report
      2021 Annual Research Report
    • Int'l Joint Research
  • [Presentation] Parametric Programming Approach for More Powerful and General Lasso Selective Inference2021

    • Author(s)
      Vo Nguyen Le Duy, Ichiro Takeuchi
    • Organizer
      International Conference on Artifical Intelligence and Statistics (AISTATS2021)
    • Related Report
      2021 Annual Research Report
    • Int'l Joint Research
  • [Presentation] Mean-Variance Analysis in Bayesian Optimization under Uncertainty2021

    • Author(s)
      Shogo Iwazaki, Yu Inatsu, Ichiro Takeuchi
    • Organizer
      International Conference on Artifical Intelligence and Statistics (AISTATS2021)
    • Related Report
      2021 Annual Research Report
    • Int'l Joint Research
  • [Presentation] More Powerful and General Selective Inference for Stepwise Feature Selection using Homotopy Method2021

    • Author(s)
      Kazuya Sugiyama, Vo Nguyen Le Duy, Ichiro Takeuchi
    • Organizer
      International Conference on Machine Learning 2021 (ICML2021)
    • Related Report
      2021 Annual Research Report
    • Int'l Joint Research
  • [Presentation] Active Learning for Distributionally Robust Level-Set Estimation2021

    • Author(s)
      Yu Inatsu, Shogo Iwazaki, Ichiro Takeuchi
    • Organizer
      International Conference on Machine Learning 2021 (ICML2021)
    • Related Report
      2021 Annual Research Report
    • Int'l Joint Research
  • [Presentation] パラメトリック計画法による選択的推論とその応用2020

    • Author(s)
      竹内一郎
    • Organizer
      電子情報通信学会IBISML研究会
    • Related Report
      2020 Annual Research Report
    • Invited
  • [Presentation] Computing Valid p-value for Optimal Changepoint by Selective Inference using Dynamic Programming.2020

    • Author(s)
      Vo Nguyen Le Duy, Hiroki Toda, Ryota Sugiyama, Ichiro Takeuchi
    • Organizer
      34th Conference on Neural Information Processing Systems (NeurIPS2020)
    • Related Report
      2020 Annual Research Report
    • Int'l Joint Research
  • [Presentation] Computing Valid P-values for Image Segmentation by Selective Inference.2020

    • Author(s)
      Kosuke Tanizaki, Noriaki Hashimoto, Yu Inatsu, Hidekata Hontani, Ichiro Takeuchi
    • Organizer
      IEEE/CVF Conference on Computer Vision and Pattern Recognition 2020 (CVPR2020)
    • Related Report
      2020 Annual Research Report
    • Int'l Joint Research
  • [Presentation] Multi-scale Domain-adversarial Multiple-instance CNN for Cancer Subtype Classification with Non-annotated Histopathological Images2020

    • Author(s)
      Noriaki Hashimoto, Daisuke Fukushima, Ryoichi Koga, Yusuke Takagi, Kaho Ko, Kei Kohno, Masato Nakaguro, Shigeo Nakamura, Hidekata Hontani, Ichiro Takeuchi
    • Organizer
      IEEE/CVF Conference on Computer Vision and Pattern Recognition 2020 (CVPR2020)
    • Related Report
      2020 Annual Research Report
    • Int'l Joint Research

URL: 

Published: 2020-04-28   Modified: 2024-12-25  

Information User Guide FAQ News Terms of Use Attribution of KAKENHI

Powered by NII kakenhi