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

2021 Fiscal Year Final Research Report

Next Optimization Methods for Social Implementation of Machine Learning Systems

Research Project

  • PDF
Project/Area Number 19H00808
Research Category

Grant-in-Aid for Scientific Research (A)

Allocation TypeSingle-year Grants
Section一般
Review Section Medium-sized Section 25:Social systems engineering, safety engineering, disaster prevention engineering, and related fields
Research InstitutionTokyo Institute of Technology

Principal Investigator

Mizuno Shinji  東京工業大学, 工学院, 教授 (90174036)

Co-Investigator(Kenkyū-buntansha) 中田 和秀  東京工業大学, 工学院, 教授 (00312984)
北原 知就  九州大学, 経済学研究院, 准教授 (10551260)
鮭川 矩義  筑波大学, システム情報系, 助教 (20757710)
高澤 陽太朗  青山学院大学, 理工学部, 助教 (20871130)
後藤 順哉  中央大学, 理工学部, 教授 (40334031)
高野 祐一  筑波大学, システム情報系, 准教授 (40602959)
Project Period (FY) 2019-04-01 – 2022-03-31
Keywords社会システム工学 / 経営工学 / オペレーションズリサーチ / 最適化 / 機械学習 / 社会実装
Outline of Final Research Achievements

In order to promote the social implementation of machine learning systems, we here focused on optimization algorithms that offer their computational foundations. We expanded the problem class that can be solved with high accuracy in reasonable time by enhancing and nicely applying conic optimization techniques. We also developed efficient approximation algorithms especially for discrete optimization problems that have wide applications. On the other hand, we developed modeling methodologies such that the resulting outputs can be easily accepted in practice by users of machine learning systems, as well as efficient algorithms for them. In particular, we found that imposing users’ knowledge as a constraint in the learning stage without compromising its performance is effective in practice.

Free Research Field

経営工学

Academic Significance and Societal Importance of the Research Achievements

Society 5.0の実現に向けては機械学習システムをより安心・安全に利用するための方法論の確立が重要であり,本研究では,その方法論の計算基盤である最適化アルゴリズムを刷新した.成果の一部はライブラリ等で公開され,比較的容易に利用できる形となっており,社会のさまざまな場面での活躍が期待できる.また,純粋数学の未解決問題に貢献する成果も得られた.事前知識の活用に基づく機械学習におけるモデリング技術は,汎用性が高く,また,高度な数学を用いずともその有効性や妥当性を議論できるものであり,一部の専門家による利用にとどまらず,こちらについても,社会のさまざまな場面での活躍が期待できる.

URL: 

Published: 2023-01-30  

Information User Guide FAQ News Terms of Use Attribution of KAKENHI

Powered by NII kakenhi