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

Scalable non-linear and non-convex manifold optimization for big data machine learning

Research Project

Project/Area Number 19K12115
Research Category

Grant-in-Aid for Scientific Research (C)

Allocation TypeMulti-year Fund
Section一般
Review Section Basic Section 61030:Intelligent informatics-related
Research InstitutionWaseda University

Principal Investigator

Kasai Hiroyuki  早稲田大学, 理工学術院, 教授 (40312079)

Project Period (FY) 2019-04-01 – 2022-03-31
Project Status Completed (Fiscal Year 2021)
Budget Amount *help
¥4,290,000 (Direct Cost: ¥3,300,000、Indirect Cost: ¥990,000)
Fiscal Year 2021: ¥1,820,000 (Direct Cost: ¥1,400,000、Indirect Cost: ¥420,000)
Fiscal Year 2020: ¥1,430,000 (Direct Cost: ¥1,100,000、Indirect Cost: ¥330,000)
Fiscal Year 2019: ¥1,040,000 (Direct Cost: ¥800,000、Indirect Cost: ¥240,000)
Keywords最適化 / 多様体 / 確率的勾配法 / 最適輸送問題 / Wasserstein距離 / リーマン多様体 / 深層学習 / 機械学習 / 多様体最適化 / ビッグデータ / 非線形非凸
Outline of Research at the Start

本研究では,スケーラブルな非線形非凸多様体最適化手法について研究する.具体的には,直交部分空間や正定値行列等の多様体制約付き深層学習のための深層学習確率的勾配法,二次最適性を実現する非厳密三次正則化法及び非厳密信頼領域,緩和更新処理に基づく確率的勾配法,またそれらの応用,オープンソフトウェアの公開と実データを用いた評価,について研究する.

Outline of Final Research Achievements

This project addresses some scalable non-linear and non-convex optimization method on manifolds for big-data machine learning problems. For this, this project particularly considers stochastic gradient algorithms on manifolds for deep learning, a second-order inexact trust-region algorithms on manifolds, and those application studies on some numerical applications. In addition, optimal transport problem is addressed, and its optimization algorithm and some applications are investigated.

Academic Significance and Societal Importance of the Research Achievements

多くの検討の対象がユークリッド空間を対象としているのに対して,本研究が対象とする空間はリーマン多様体であることから,その間に大きな理論上・実装上・実応用上の隔たりがある.リーマン多様体最適化は,従来のユークリッド空間における制約付き最適化とは全く思想が異なり,近年研究が本格化された極めて新しい手法である.日本国内での理論的・実践的研究例は極めて少ない.本研究は,これまでの応募者の研究成果を発展させ,近年進展が著しい深層学習についてリーマン多様体最適化の視点からアプローチするものであり,他に例を見ない極めて斬新な取り組みである.

Report

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

    (23 results)

All 2021 2020 2019 Other

All Int'l Joint Research (1 results) Journal Article (10 results) (of which Int'l Joint Research: 2 results,  Peer Reviewed: 10 results,  Open Access: 2 results) Presentation (12 results) (of which Invited: 1 results)

  • [Int'l Joint Research] マイクロソフト(インド)

    • Related Report
      2019 Research-status Report
  • [Journal Article] Block-coordinate Frank-Wolfe algorithm and convergence analysis for semi-relaxed OT problem2021

    • Author(s)
      Takumi Fukunaga and Hiroyuki Kasai
    • Journal Title

      IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

      Volume: NA Pages: 5433-5437

    • DOI

      10.1109/icassp43922.2022.9746032

    • Related Report
      2021 Annual Research Report
    • Peer Reviewed
  • [Journal Article] LCS graph kernel based on Wasserstein distance in longest common subsequence metric space2021

    • Author(s)
      JianmingHuang, Zhongxi Fang, and HiroyukiKasai
    • Journal Title

      Signal Processing

      Volume: 189 Pages: 1-13

    • DOI

      10.1016/j.sigpro.2021.108281

    • Related Report
      2021 Annual Research Report
    • Peer Reviewed
  • [Journal Article] Graph embedding using multi-layer adjacent point merging model2021

    • Author(s)
      J. Huang and H. Kasai
    • Journal Title

      IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

      Volume: 0 Pages: 0-0

    • Related Report
      2020 Research-status Report
    • Peer Reviewed
  • [Journal Article] Riemannian optimization on the simplex of positive definite matrices2020

    • Author(s)
      B. Mishra, H. Kasai, and P. Jawanpuria
    • Journal Title

      NeurIPS workshop OPT2020

      Volume: 0 Pages: 0-0

    • Related Report
      2020 Research-status Report
    • Peer Reviewed / Open Access / Int'l Joint Research
  • [Journal Article] Consistency-aware and inconsistency-aware graph-based multi-view clustering2020

    • Author(s)
      M. Horie and H. Kasai
    • Journal Title

      28th International European Signal Processing Conference (EUSIPCO) 2020

      Volume: 0 Pages: 0-0

    • DOI

      10.23919/eusipco47968.2020.9287516

    • Related Report
      2020 Research-status Report
    • Peer Reviewed
  • [Journal Article] Wasserstein k-means with sparse simplex projection2020

    • Author(s)
      T. Fukunaga and H. Kasai
    • Journal Title

      25th International Conference on Pattern Recognition (ICPR) 2020

      Volume: 0 Pages: 0-0

    • DOI

      10.1109/icpr48806.2021.9412131

    • Related Report
      2020 Research-status Report
    • Peer Reviewed
  • [Journal Article] Multi-view Wasserstein discriminant analysis with entropic regularized Wasserstein distance2020

    • Author(s)
      Hiroyuki Kasai
    • Journal Title

      IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

      Volume: NA Pages: 1-5

    • DOI

      10.1109/icassp40776.2020.9054427

    • NAID

      40022116900

    • Related Report
      2019 Research-status Report
    • Peer Reviewed
  • [Journal Article] Sequential Semi-Orthogonal Multi-Level NMF with Negative Residual Reduction for Network Embedding2020

    • Author(s)
      Riku Hashimoto, Hiroyuki Kasai
    • Journal Title

      IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

      Volume: NA Pages: 1-5

    • DOI

      10.1109/icassp40776.2020.9054660

    • NAID

      40022116893

    • Related Report
      2019 Research-status Report
    • Peer Reviewed
  • [Journal Article] Overlapped State Hidden Semi-Markov Model for Grouped Multiple Sequences2020

    • Author(s)
      Hiromi Narimatsu, Hiroyuki Kasai
    • Journal Title

      IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

      Volume: NA Pages: 1-5

    • DOI

      10.1109/icassp40776.2020.9054136

    • Related Report
      2019 Research-status Report
    • Peer Reviewed
  • [Journal Article] Riemannian adaptive stochastic gradient algorithms on matrix manifolds2019

    • Author(s)
      Hiroyuki Kasai, Pratik Kumar Jawanpuria, Bamdev Mishra
    • Journal Title

      36th International Conference on Machine Learning (ICML2019)

      Volume: PMLR 97 Pages: 3262-3271

    • Related Report
      2019 Research-status Report
    • Peer Reviewed / Open Access / Int'l Joint Research
  • [Presentation] エントロピー正則付き半緩和最適輸送問題におけるSinkhornアルゴリズムに関する検討2021

    • Author(s)
      福永 拓海, 笠井 裕之
    • Organizer
      情報処理学会AVM研究会
    • Related Report
      2021 Annual Research Report
  • [Presentation] 隣接要素関係性の自動重み付けによる最適輸送シーケンスマッチング2021

    • Author(s)
      堀江 光彦, 笠井 裕之
    • Organizer
      情報処理学会AVM研究会
    • Related Report
      2021 Annual Research Report
  • [Presentation] 嘉村 雅志, 笠井 裕之2021

    • Author(s)
      大域的な順序変更に対応するDTWを用いたシーケンスマッチング手法の検討
    • Organizer
      人工知能学会全国大会論文集
    • Related Report
      2021 Annual Research Report
  • [Presentation] 質量制御最適輸送によるグラフ構造比較手法2021

    • Author(s)
      方 鐘熙, 黄 健明, 笠井 裕之
    • Organizer
      人工知能学会全国大会論文集
    • Related Report
      2021 Annual Research Report
  • [Presentation] Frank-Wolfeアルゴリズムに基づく緩和最適輸送問題のための高速最適化手法の検討2021

    • Author(s)
      福永 拓海, 笠井 裕之
    • Organizer
      人工知能学会全国大会論文集
    • Related Report
      2021 Annual Research Report
  • [Presentation] 緩和最適輸送問題のための Frank-Wolfeアルゴリズム高速化手法と色転写問題への応用2021

    • Author(s)
      福永拓海,笠井裕之
    • Organizer
      情報処理学会AVM研究会
    • Related Report
      2020 Research-status Report
  • [Presentation] 隣接要素の関連と時間分布の違いを考慮したグループ化シーケンスマッチング手法の検討2021

    • Author(s)
      堀江光彦,笠井裕之
    • Organizer
      情報処理学会AVM研究会
    • Related Report
      2020 Research-status Report
  • [Presentation] OT-based graph similarity using shortest-path length,2020

    • Author(s)
      Z. Fang and H. Kasai
    • Organizer
      情報処理学会AVM研究会
    • Related Report
      2020 Research-status Report
  • [Presentation] A Wasserstein graph kernel based on substructure isomorphism problem of shortest paths2020

    • Author(s)
      J. Huang and H. Kasai
    • Organizer
      情報処理学会AVM研究会
    • Related Report
      2020 Research-status Report
  • [Presentation] ネットワーク・ワードエンベディングのための負値残差低減および半直交制約付き非負値行列分解2019

    • Author(s)
      橋本陸 ,笠井裕之
    • Organizer
      研究報告オーディオビジュアル複合情報処理(AVM)
    • Related Report
      2019 Research-status Report
  • [Presentation] エントロピー正則化Wasserstein距離に基づくマルチビューWasserstein判別法2019

    • Author(s)
      笠井裕之
    • Organizer
      研究報告オーディオビジュアル複合情報処理(AVM)
    • Related Report
      2019 Research-status Report
  • [Presentation] 非線形構造制約付き大規模機械学習問題への取り組み ~非厳密リーマン多様体最適化の理論と応用~2019

    • Author(s)
      笠井裕之
    • Organizer
      研究報告オーディオビジュアル複合情報処理(AVM)
    • Related Report
      2019 Research-status Report
    • Invited

URL: 

Published: 2019-04-18   Modified: 2023-01-30  

Information User Guide FAQ News Terms of Use Attribution of KAKENHI

Powered by NII kakenhi