Hiroshi Kera

Education

• 2020.03

  PhD, Department of Electrical Engineering and Information Systems, The University of Tokyo

  Doctoral thesis: "Learning Vanishing Ideals under Noise"

• 2017.03

  Master's, Department of Electrical Engineering and Information Systems, The University of Tokyo

  Master's thesis: "Detecting Objects of Shared Attention from First-Person Videos"

• 2015.03

  BEng, Department of Electronic and Electrical Engineering, The University of Tokyo

Positions

• 2025.10 – Present

  Project Researcher, National Institute of Informatics

• 2025.04 – Present

  Associate Professor, Institute for Advanced Academic Research / Graduate School of Informatics, Chiba University

• 2024.10 – 2025.09

  Guest Researcher, Zuse Institute Berlin

• 2024.04 – 2025.03

  Assistant Professor, Graduate School of Informatics, Chiba University

• 2022.07 – 2022.08

  Technical Advisor, Fujitsu Limited

• 2022.04 – 2023.03

  Technical Advisor, Ricoh Company, Ltd.

• 2021.06 – 2022.05

  Visiting Researcher, The University of Tokyo

• 2021.04 – 2024.03

  Assistant Professor, Graduate School of Engineering, Chiba University

• 2020.08 – 2020.10

  Visiting Researcher, AIP-NEC Collaboration Center, RIKEN Center for Advanced Intelligence Project

• 2020.04 – 2021.03

  Project Researcher, The University of Tokyo

• 2017.04 – 2020.03

  Research Fellow (DC1), Japan Society for the Promotion of Science

Grants

• Maximizing the Utility of Chain-of-Thought in Autoregressive GenerationPI

  Grants-in-Aid for Scientific Research / Grant-in-Aid for Scientific Research (B) / Japan Society for the Promotion of Science

  2026-04 – 2030-03 · Project direct costs: ¥14,200,000

• 個別最適化AIを実現する軽量な転移学習法の確立―算術的な知識編集法の開発―Co-Investigator

  科学研究費助成事業 / 基盤研究(B) / 日本学術振興会

  2026-04 – 2030-03 · Project direct costs: ¥19,000,000

• Zero-Shot Textual Explanation of Decision Rationale in Image Recognition Models

  Open Collaborative Research / National Institute of Informatics

  2026-04 – 2027-03

• 深層モデルの入力応答特性の包括的制御PI

  BOOST 次世代AI人材育成プログラム（若手研究者支援） / 国立研究開発法人科学技術振興機構

  2025-10 – 2030-09 · Project direct costs: ¥50,000,000

• Transforming Weather Prediction with Deep Learning and Quantum ComputingCo-Investigator

  Grants-in-Aid for Scientific Research / Grant-in-Aid for Scientific Research (A) / Japan Society for the Promotion of Science

  2025-04 – 2030-03 · Project direct costs: ¥35,100,000

• Improving AI Quality through Deep Model Integration

  Open Collaborative Research / National Institute of Informatics

  2025-04 – 2026-03

• ドメイン適応に関する学術指導PI

  JFEエンジニアリング株式会社

  2024-10 – 2025-03

• 深層計算代数の開拓PI

  研究支援プログラム（J-PEAKS） / 千葉大学国際高等研究基幹（IAAR）

  2024-10 – 2026-09 · Project direct costs: ¥5,000,000

• 深層消失関数の計算と理論の深化PI

  科学研究費助成事業 / 国際共同研究加速基金(国際共同研究強化) / 日本学術振興会

  2024-08 – 2027-03 · Project direct costs: ¥9,200,000

• End-to-end 自動運転モデルの研究開発Co-Investigator

  Suzuca Technologies 株式会社

  2024-07 – 2026-08

• 機械学習代数攻撃の実行可能性とその対策法PI

  三菱電機株式会社

  2024-04 – 2027-03

• 撮影条件にロバストな特徴マッチング技術の開発PI

  リコージャパン株式会社

  2024-03 – 2026-03

• 大規模代数計算のための計算代数および機械学習理論の構築PI

  戦略的な研究開発の推進 戦略的創造研究推進事業 さきがけ / 科学技術振興機構

  2024 – 2028-03 · Project direct costs: ¥30,000,000

• Artificial generation of upstream maritime heavy rains to govern intense-rain-induced disasters over land (AMAGOI)Co-Investigator

  MOONSHOT / Japan Science and Technology Agency

  2023-12 – 2027-03

  Lead of subtask 2-2, development of weather prediction surrogate models with attention mechanisms. Direct costs allocated to Kera: JPY 34,990,000 in total for FY2023–FY2026.

• 大規模代数方程式系の学習と処理PI

  戦略的な研究開発の推進 戦略的創造研究推進事業 ACT-X / 科学技術振興機構

  2023-10 – 2025-03 · Project direct costs: ¥4,500,000

• 歩行者軌道予測モデル最適化のための予測対象歩行者の速度・加速度の推定Co-Investigator

  京セラ株式会社

  2023-10 – 2024-09

• ロバストIoT×AIのための深層防御システムの研究調査Co-Investigator

  研究調査助成 / 電気通信普及財団

  2023-10 – 2024-09 · Project direct costs: ¥1,755,000

• 深層ゼロ関数学習の計算と理論PI

  科学研究費助成事業 / 若手研究 / 日本学術振興会

  2022-04 – 2025-03 · Project direct costs: ¥3,500,000

• Adversarial design in deep learningCo-Investigator

  Grants-in-Aid for Scientific Research / Grant-in-Aid for Scientific Research (B) / Japan Society for the Promotion of Science

  2022-04 – 2026-03 · Project direct costs: ¥13,200,000

• 画像センシングによる重機操作における安全検知技術の開発Co-Investigator

  住友建機株式会社

  2021-10 – 2026-09

• データドリブン計算代数幾何PI

  戦略的な研究開発の推進 戦略的創造研究推進事業 ACT-X / 科学技術振興機構

  2020-10 – 2022-03 · Project direct costs: ¥4,500,000

• Theory and Application of deep learning models through the lens of computational algebraic geometryPI

  Grants-in-Aid for Scientific Research / Grant-in-Aid for Research Activity Start-up / Japan Society for the Promotion of Science

  2020-09 – 2022-03 · Project direct costs: ¥2,200,000

• SNSにおける人気指標向上及び投稿内容に関する研究Co-Investigator

  株式会社サイバー・バズ

  2020-04 – 2021-03

• 広告クリエイティブにおける効果予測及び魅力度解析に関する研究Co-Investigator

  株式会社セプテーニ

  2020-04 – 2021-03

• 外れ値への頑健性および多目的最適化を導入した近似計算代数アルゴリズムの開発PI

  科学研究費助成事業 特別研究員奨励費 / 特別研究員奨励費 / 日本学術振興会

  2017-04 – 2020-03 · Project direct costs: ¥2,800,000

Selected Awards

• Silver Reviewer Award

  International Conference on Machine Learning (ICML) 2026 · 2026-07

• Top Review Award

  Thirty-Ninth Annual Conference on Neural Information Processing Systems (NeurIPS) · 2025-12

• MIRU論文評価貢献賞

  CVIM研究会 · 2025-08

• 電子情報学専攻長賞

  東京大学大学院 情報理工学系研究科 電子情報学専攻 · 2020-03

• IEEE CIS Japan Chapter Young Researcher Award (Evolutionary Computation)

  進化計算学会 · 2015-12

• 優秀卒業論文賞

  東京大学 工学部 電子情報工学科 · 2015-03

Selected Publications

1. Zero-Shot Textual Explanations via Translating Decision-Critical Features

   Toshinori Yamauchi, Hiroshi Kera, Kazuhiko Kawamoto

   Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Findings, 2026 · pp. 3282–3292

   PaperarXiv

2. ★ Explaining Object Detectors via Collective Contribution of Pixels

   Toshinori Yamauchi, Hiroshi Kera, Kazuhiko Kawamoto

   Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2026 · pp. 17046–17056

   PaperarXiv

3. ★ Adversarially Pretrained Transformers May Be Universally Robust In-Context Learners

   Soichiro Kumano, Hiroshi Kera, Toshihiko Yamasaki

   International Conference on Learning Representations (ICLR), 2026 · pp. 124615–124659

   PaperarXiv

4. ★ Matching Semantically Similar Non-Identical Objects

   Yusuke Marumo, Kazuhiko Kawamoto, Satomi Tanaka, Shigenobu Hirano, Hiroshi Kera*

   IEEE/CVF Winter Conference on Applications of Computer Vision (WACV), 2026 · pp. 2752–2764

   PaperarXiv

5. ★ Computational Algebra with Attention: Transformer Oracles for Border Basis Algorithms

   Hiroshi Kera*†, Nico Pelleriti†, Yuki Ishihara, Max Zimmer, Sebastian Pokutta

   Advances in Neural Information Processing Systems (NeurIPS), 2025 · pp. 9438–9480

   DOIPaperarXiv

6. ★ Adapter Merging with Centroid Prototype Mapping for Scalable Class-Incremental Learning.

   Takuma Fukuda, Hiroshi Kera, Kazuhiko Kawamoto

   IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2025 · pp. 4884–4893

   DOIPaperarXiv

7. ★ Learning to compute Gröbner bases

   Hiroshi Kera*, Yuki Ishihara, Yuta Kambe, Tristan Vaccon, Kazuhiro Yokoyama

   Advances in Neural Information Processing Systems (NeurIPS), 2024 · pp. 33141–33187

   DOIPaperarXiv

8. ★ Wide Two-Layer Networks can Learn from Adversarial Perturbations

   Soichiro Kumano, Hiroshi Kera, Toshihiko Yamasaki

   Advances in Neural Information Processing Systems, 2024 · pp. 59755–59807

   DOIPaperarXiv

9. Monomial-agnostic computation of vanishing ideals

   Hiroshi Kera*, Yoshihiko Hasegawa

   Journal of Computational Algebra, 2024 · pp. 100022

   DOIarXiv

10. ★ Identifying Important Group of Pixels using Interactions

    Kosuke Sumiyasu, Kazuhiko Kawamoto, Hiroshi Kera*

    Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2024 · pp. 6017–6026

    PaperarXiv

11. Fourier analysis on robustness of graph convolutional neural networks for skeleton-based action recognition

    Nariki Tanaka, Hiroshi Kera, Kazuhiko Kawamoto

    Computer Vision and Image Understanding, 2024 · pp. 103936–103936

    DOIarXiv

12. ★ Adversarial Training from Mean Field PerspectiveSpotlight

    Soichiro Kumano, Hiroshi Kera, Toshihiko Yamasaki

    Advances in Neural Information Processing Systems, 2023 · pp. 75097–75150

    DOIPaperarXiv

13. ★ Approximate Vanishing Ideal Computations at Scale.

    Elias Samuel Wirth, Hiroshi Kera, Sebastian Pokutta

    International Conference on Learning Representations, 2023

    PaperarXiv

14. ★ Border Basis Computation with Gradient-Weighted Normalization.

    Hiroshi Kera*

    International Symposium on Symbolic and Algebraic Computation (ISSAC), 2022 · pp. 225–234

    DOIarXiv

15. ★ Adversarial Bone Length Attack on Action Recognition

    Nariki Tanaka, Hiroshi Kera, Kazuhiko Kawamoto

    Proceedings of the AAAI Conference on Artificial Intelligence, 2022, 2022 · pp. 2335–2343

    DOIarXiv

16. ★ Gradient Boosts the Approximate Vanishing Ideal.

    Hiroshi Kera, Yoshihiko Hasegawa

    The Thirty-Fourth AAAI Conference on Artificial Intelligence(AAAI), 2020 · pp. 4428–4435

    DOIPaperarXiv

17. Spurious Vanishing Problem in Approximate Vanishing Ideal

    Hiroshi Kera, Yoshihiko Hasegawa

    IEEE Access, 2019 · pp. 178961–178976

    DOIarXiv

18. ★ Approximate Vanishing Ideal via Data Knotting

    Hiroshi Kera, Yoshihiko Hasegawa

    Proceedings of the AAAI Conference on Artificial Intelligence, 2018 · pp. 3399–3406

    DOIarXiv

19. Noise-tolerant algebraic method for reconstruction of nonlinear dynamical systems

    Hiroshi Kera, Yoshihiko Hasegawa

    Nonlinear Dynamics, 2016 · pp. 675–692

    DOIPaper

Academic Service

Program Committees and Reviewing

• NeurIPS (Conference on Neural Information Processing Systems) — 2024, 2025, 2026
• ICLR (International Conference on Learning Representations) — 2024, 2025, 2026
• ICML (International Conference on Machine Learning) — 2025, 2026
• IJCAI (International Joint Conference on Artificial Intelligence) — 2021 (Senior), 2022, 2023, 2024
• CVPR (IEEE/CVF Computer Vision and Pattern Recognition) — 2023 (external reviewer), 2024, 2025
• ISSAC (International Symposium on Symbolic and Algebraic Computation) — 2026 (external reviewer)
• ICME (IEEE International Conference on Multimedia & Expo) — 2021, 2022, 2023
• MIPR (IEEE International Conference on Multimedia Information Processing and Retrieval) — 2021, 2022
• ACM Multimedia — 2021
• ICMR (ACM International Conference on Multimedia Retrieval) — 2021
• MIRU (Meeting on Image Recognition and Understanding) — 2021, 2022, 2023, 2024, 2025

Organizing / Steering Committee

• Steering Committee, Symbolic Computation and Machine Learning (SCML) (2025.07–)
• Organizing Committee, Kyushu University IMI Joint Research "Accelerating Symbolic Computation and Its Application to Industrial Problems" (2023, 2024, 2025)
• Organizing Committee, RIMS Joint Research "Deepening and Applications of Computational Algebraic Geometry via Gröbner Bases" (2025)
• Tutorial Moderator, MIRU 2022

Society Memberships

• IEEE Computational Intelligence Society (2016.11–)
• Japan Society for Symbolic and Algebraic Computation (2023.05–)
• The Japanese Society for Artificial Intelligence (2025.07–)

Research

• Learning Computational Patterns — Arithmetics, Algebra, and More
• Explaining Vision Models via Game-Theoretic Indices
• Non-Identical Object Matching
• Weather Modification for Heavy-Rainfall Mitigation
• Approximate Vanishing Ideals
• Human-imperceptible Perturbations