Daiki Nishiyama

I am a Ph.D. student at the School of Computing, Institute of Science Tokyo (former Tokyo Institute of Technology).

I belong to the Jun Sakuma Laboratory and engage in research on explainable AI, AI security, domain generalization, and other areas to ensure the reliability of machine learning models.

Education

Institute of Science Tokyo (Apr 2023 - Present, former Tokyo Institute of Technology): Doctoral program, Graduate Major in Artificial Intelligence, Department of Computer Science, School of Computing. Academy of Super Smart Society.
University of Tsukuba (Apr 2021 - Mar 2023): Master of Engineering, Degree Programs in Systems and Information Engineering, Graduate School of Science and Technology.
University of Tsukuba (Apr 2017 - Mar 2021): Bachelor of Engineering, College of Information Science, Major in Intelligent Information Media.

Professional Experience

Japan Society for the Promotion of Science (Apr 2024 - Present): Research Fellow DC2
Ministry of Defense, Japan Self-Defense Forces (Jul 2023 - Present): Reserve Officer (System Protection)
RIKEN, Center for Advanced Intelligence Project (AIP) (May 2021 - Present): Part-time worker I
TelHi Inc. (Jan 2019 - Present): Head of Creative Business Division (Currently on leave)
NEC Corporation (Aug 2023 - Sep 2023): Research Intern, Secure System Platform Research Laboratories
University of Tsukuba (Apr 2023 - Aug 2023): Technical Assistant
CYBERDYNE Inc. (Oct 2020 - Mar 2023): Web Developer
AVILEN Inc. (Jul 2020 - Mar 2024): Internship
Feedal Inc. (Apr 2019 - Apr 2021): Director, Chief Experience Officer (CXO)

Publications

Journal

Daiki Nishiyama, Kazuto Fukuchi, Youhei Akimoto, and Jun Sakuma. “CAMRI Loss: Improving the Recall of a Specific Class without Sacrificing Accuracy.” IEICE Transactions on Information and Systems, Vol.E106-D, No.4, pp.523-537, April 2023. [link]

Conference

Daiki Nishiyama, Hiroaki Miyoshi, Noriaki Hashimoto, Koichi Ohshima, Hidekata Hontani, Ichiro Takeuchi, and Jun Sakuma. Explainable classifier for malignant lymphoma subtyping via cell graph and image fusion. Proceedings of Medical Image Computing and Computer Assisted Intervention – MICCAI 2025, LNCS 15971, pages 320–330. Springer Nature Switzerland, September 2025. [link]
Daiki Nishiyama, Kazuto Fukuchi, Youhei Akimoto, and Jun Sakuma, “CAMRI Loss: Improving Recall of a Specific Class without Sacrificing Accuracy.” 2022 International Joint Conference on Neural Networks (IJCNN), IEEE, No. 503, July 2022. [link]

Conference (in Japanese)

北野尚樹, 西山大輝．「法令文の可読性向上のための定義規定・略称規定における文型定義及びパターンベースの正式名称・略称抽出手法」, 言語処理学会第31回年次大会 (NLP2025), B6-5, pp.2243-2248, 長崎, 2025年3月. [link]
永池礼, 西山大輝, 秋本洋平, 福地一斗, 佐久間淳．「マルチタスク学習における隠れたタスクに対する敵対的攻撃」2024年暗号と情報セキュリティシンポジウム（SCIS2024), 電子情報通信学会情報セキュリティ研究専門委員会（ISEC研）, 2024. 1. 26．
西山大輝，福地一斗，秋本洋平，佐久間淳．「精度劣化を伴わない特定クラスの再現率改善のための分類器学習」第37回人工知能学会全国大会, 人工知能学会, No. 3D1-GS-2-03, 熊本, 2023年6月. [link]
西山大輝, 福地一斗, 秋本洋平, 佐久間淳．「特定クラスのリコール改善のための加法的角度マージン損失による分類器学習」第44回情報論的学習理論と機械学習研究会, 電子情報通信学会, No. 5, オンライン, 2022年1月. [link]
西山大輝, 福地一斗, 秋本洋平, 佐久間淳． “Class-Sensitive Loss: Improving the Recall of a Particular Class without Sacrificing Accuracy.” 第24回情報論的学習理論ワークショップ, 電子情報通信学会, No. 100, オンライン, 2021年11月.

Invited Talks

西山大輝，三好寛明，橋本典明，大島孝一，本谷秀堅，竹内一郎，佐久間淳．「悪性リンパ腫の病理画像分類における細胞に基づく統計的特徴解析と説明可能な分類モデル」第6回日本メディカルAI学会学術集会，2024年6月．

Takeuchi, Jun Sakuma,)

Preprint

Daiki Nishiyama, Hiroaki Miyoshi, Noriaki Hashimoto, Koichi Ohshima, Hidekata Hontani, Ichiro Takeuchi and Jun Sakuma. “Explainable Classifier for Malignant Lymphoma Subtyping via Cell Graph and Image Fusion.” arXiv preprint arXiv:2503.00925 (2025). [link]
Yu Zhe, Rei Nagaike, Daiki Nishiyama, Kazuto Fukuchi, Jun Sakuma, “Adversarial Attacks on Hidden Tasks in Multi-Task Learning.” arXiv preprint arXiv:2405.15244, 2024. [link]
Daiki Nishiyama, Kazuto Fukuchi, Youhei Akimoto, Jun Sakuma, “CAMRI Loss: Improving Recall of a Specific Class without Sacrificing Accuracy.” arXiv preprint arXiv:2209.10920, 2022. [link]

Research Projects

Daiki Nishiyama, “Reliable Classification Model for Explaining Relationships between Objects in Images”. Japan Society for the Promotion of Science, Research Fellowship for Young Scientists 24KJ1049, 2024-2026. [link]

Awards

Tokyo Tech Tsubame Scholarship for Doctoral Students (2023)
Program Leader Award, Degree Programs in Systems and Information Engineering, University of Tsukuba (2023)
Special Program Leader Award, Degree Programs in Systems and Information Engineering, University of Tsukuba (2022)
enPiT Excellence Award, Joint Presentation by University of Tsukuba and University of the Ryukyus (2021)
Program Leader Award, College of Information Science, University of Tsukuba (2021)
Winner, 1st TDK Presents Student Innovation Battle (2020)
Best Award, T-ACT 2019 First Half (2019)

Skills and Certifications

Registered Information Security Specialist (2023)
Middle School Teaching License (Mathematics and Information)
High School Teaching License (Mathematics and Information)
CG Engineer Certification (Expert) (2018)

Contact Information

Email: nishiyama.d.2d7f[at]m.isct.ac.jp
Address: E-904, West Building 8, Institute of Science Tokyo, 2-12-1 Ookayama, Meguro-ku, Tokyo 152-8550, Japan