Updated on 2025/10/04

写真a

 
大石 慶一朗
 
Organization
Faculty of Environmental, Life, Natural Science and Technology Assistant Professor
Position
Assistant Professor
External link

Degree

  • Ph.D. in Engineering ( 2025.3   The University of Electro-Communications )

Education

  • The University of Electro-Communications   大学院情報理工学研究科   情報学専攻 博士後期課程

    2022.4 - 2025.3

      More details

  • The University of Electro-Communications   大学院情報理工学研究科   情報学専攻 博士前期課程

    2017.4 - 2019.3

      More details

  • The University of Electro-Communications   情報理工学部   総合情報学科

    2013.4 - 2017.3

      More details

Research History

  • Okayama University   Faculty of Environmental, Life, Natural Science and Technology   Assistant Professor

    2025.4

      More details

  • 富士通株式会社

    2019.4 - 2020.12

      More details

Professional Memberships

Committee Memberships

  • 電子情報通信学会 人工知能と知識処理 専門委員会   幹事補佐  

    2025.7   

      More details

    Committee type:Academic society

    researchmap

 

Papers

  • Requirements Analysis of Federated Learning through Context Engineering Reviewed

    Keiichiro Oishi, Shinichi Honiden, Hiroyuki Nakagawa

    2025.11

     More details

    Authorship:Lead author, Corresponding author   Publishing type:Research paper (conference, symposium, etc.)  

    researchmap

  • Federated Learning Algorithm to Suppress Occurrence of Low-Accuracy Devices Reviewed International coauthorship International journal

    Koudai Sakaida, Keiichiro Oishi, Yasuyuki Tahara, Akihiko Ohsuga, Andrew J, Yuichi Sei

    International journal of electrical and computer engineering systems   16 ( 8 )   607 - 620   2025.9

     More details

    Publishing type:Research paper (scientific journal)   Publisher:Faculty of Electrical Engineering, Computer Science and Information Technology Osijek  

    In recent years, federated learning (FL), a decentralized machine learning approach, has garnered significant attention. FL enables multiple devices to collaboratively train a model without sharing their data. However, when the data across devices are non- independent and identically distributed (non-IID), performance degradation issues such as reduced accuracy, slower convergence speed, and decreased performance fairness are known to occur. Under non-IID data environments, the trained model tends to exhibit varying accuracies across different devices, often overfitting on some devices while achieving lower accuracy on others. To address these challenges, this study proposes a novel approach that integrates reinforcement learning into FL under Non-IID conditions. By employing a reinforcement learning agent to select the optimal devices in each round, the proposed method effectively suppresses the emergence of low-accuracy devices compared to existing methods. Specifically, the proposed method improved the average accuracy of the bottom 10% devices by up to 4%, without compromising the overall average accuracy. Furthermore, the device selection patterns revealed that devices with more diverse local data tend to be chosen more frequently.

    DOI: 10.32985/ijeces.16.8.4

    researchmap

  • Algorithm to satisfy l‐diversity by combining dummy records and grouping Reviewed International coauthorship International journal

    Keiichiro Oishi, Yuichi Sei, J. Andrew, Yasuyuki Tahara, Akihiko Ohsuga

    SECURITY AND PRIVACY   7 ( 3 )   e373:1 - e373:15   2024.5

     More details

    Authorship:Lead author, Corresponding author   Language:English   Publishing type:Research paper (scientific journal)  

    DOI: 10.1002/spy2.373

    researchmap

  • Federated Learning Algorithm Handling Missing Attributes Reviewed International journal

    Keiichiro Oishi, Yuichi Sei, Yasuyuki Tahara, Akihiko Ohsuga

    Proceedings of 6th IEEE International Conference on Internet of Things and Intelligence Systems (IoTaIS)   2023.11

     More details

    Authorship:Lead author, Corresponding author   Language:English   Publishing type:Research paper (international conference proceedings)  

    researchmap

  • Semantic diversity: Privacy considering distance between values of sensitive attribute Reviewed International journal

    Keiichiro Oishi, Yuichi Sei, Yasuyuki Tahara, Akihiko Ohsuga

    Computers & Security   94 ( 101823 )   1 - 18   2020.7

     More details

    Authorship:Lead author, Corresponding author   Language:English   Publishing type:Research paper (scientific journal)  

    DOI: 10.1016/j.cose.2020.101823

    researchmap

  • Proposal of l-diversity algorithm considering distance between sensitive attribute values Reviewed International journal

    Keiichiro Oishi, Yasuyuki Tahara, Yuichi Sei, Akihiko Ohsuga

    2017 IEEE Symposium Series on Computational Intelligence, SSCI 2017 - Proceedings   2018-   1 - 8   2018.2

     More details

    Authorship:Lead author, Corresponding author   Language:English   Publishing type:Research paper (international conference proceedings)  

    researchmap

▼display all

Books

  • Toward a secure metaverse: crafting cutting-edge algorithm for protected data analysis Reviewed International journal

    Keiichiro Oishi, Yasuyuki Tahara, Akihiko Ohsuga, Andrew J., Agbotiname Lucky Imoize, Yuichi Sei( Role: Contributor ,  Advanced Metaverse Wireless Communication Systems, Chapter 8, pp.181–207)

    The Institution of Engineering and Technology (IET)  2025.1 

     More details

    Book type:Scholarly book

    DOI: 10.1049/PBTE112E_ch8

    researchmap

MISC

  • 非独立同一分布データにおける連合学習の性能公平性改善に関する研究

    境田晃大, 大石慶一朗, 田原康之, 大須賀昭彦, 清雄一

    SMASH24 Summer Symposium   2024.9

     More details

  • 属性の欠損に対応した連合学習アルゴリズム

    大石慶一朗, 福田賢一郎, 清雄一, 田原康之, 大須賀昭彦

    SMASH24 Summer Symposium   2024.9

     More details

    Authorship:Lead author  

    researchmap

  • l-多様性を満たすためのグルーピングとダミー追加を組み合わせたアルゴリズム

    大石慶一朗, 清雄一, 田原康之, 大須賀昭彦

    電子情報通信学会 人工知能と知識処理研究会   122 ( 322 )   80 - 86   2022.12

     More details

    Authorship:Lead author  

    researchmap

  • センシティブ属性値の距離を考慮したダミー追加によるl-多様性アルゴリズムの提案

    大石慶一朗, 清雄一, 田原康之, 大須賀昭彦

    コンピュータセキュリティシンポ ジウム(CSS)   2017.10

     More details

    Authorship:Lead author  

    researchmap

  • センシティブ属性値の距離を考慮したダミー追加によるl-多様性アルゴリズムの提案

    大石, 慶一朗, 清, 雄一, 田原, 康之, 大須賀, 昭彦

    情報処理学会全国大会(第79回)   2017.3

     More details

    Authorship:Lead author  

    researchmap

Presentations

  • A Run-time Verification Toolkit for Self-adaptive Systems Supporting Dynamic Model Changes International conference

    Ryuichi Iida, Keiichiro Oishi, Hiroyuki Nakagawa

    The IEEE International Conference on Autonomic Computing and Self-Organizing Systems (ACSOS 2025)  2025.10 

     More details

    Language:English   Presentation type:Poster presentation  

    researchmap

  • Federated Learning Algorithm to Suppress Occurrence of Low-Accuracy Devices International conference

    Kodai Sakaida, Keiichiro Oishi, Yasuyuki Tahara, Akihiko Ohsuga, Yuichi Sei

    17th International Conference on Agents and Artificial Intelligence (ICAART)  2025.2 

     More details

    Presentation type:Oral presentation (general)  

    researchmap

Awards

  • Best Poster/Demo Award

    2025.10   The IEEE International Conference on Autonomic Computing and Self-Organizing Systems (ACSOS 2025)   A Run-time Verification Toolkit for Self-adaptive Systems Supporting Dynamic Model Changes

    Ryuichi Iida, Keiichiro Oishi, Hiroyuki Nakagawa

     More details

    Award type:Award from international society, conference, symposium, etc. 

    researchmap

  • 2025年度(令和7年度)山下記念研究賞

    2025.9   情報処理学会   属性の欠損に対応した連合学習アルゴリズム

    大石慶一朗

     More details

    受賞対象論文:
    大石慶一朗,福田賢一郎,清雄一,田原康之,大須賀昭彦: 属性の欠損に対応した連合学習アルゴリズム, SMASH24 Summer Symposium, 情報処理学会研究報告, Vol.2024-ICS-215, No.2, pp.1-6 (2024.9)

    researchmap

  • IEEE Computer Society Japan Chapter SMASH Young Researcher Award

    2024.9   IEEE Computer Society Tokyo/Japan Joint Chapter  

     More details

  • Best Paper Award

    2023.11   IEEE International Conference on Internet of Things and Intelligence Systems (IoTaIS)   Federated Learning Algorithm Handling Missing Attributes

     More details

  • 研究奨励賞

    2022.12   電子情報通信学会「人工知能と知識処理」研究会   l-多様性を満たすためのグルーピングとダミー追加を組み合わせたアルゴリズム

    大石慶一朗, 清雄一, 田原康之, 大須賀昭彦

     More details

Research Projects

  • 不整合データに対応したプライバシ強化連合学習技術に関する研究

    Grant number:25K24390  2025.07 - 2027.03

    日本学術振興会  科学研究費助成事業  研究活動スタート支援

    大石 慶一朗

      More details

    Grant amount:\2730000 ( Direct expense: \2100000 、 Indirect expense:\630000 )

    researchmap