Kosuke Ukita
PhD student (D1), full-time
Kyushu Institute of Technology
Education
PhD in CSE
Kyushu Institute of TechnologyDepartment of Creative Informatics, Graduate School of Computer Science and Systems Engineering
MEng in CSE
Kyushu Institute of TechnologyDepartment of Creative Informatics, Graduate School of Computer Science and Systems Engineering
Thesis: "A Study on Sensor Foundation Models via Joint Training using Flow Matching"GPA: 3.6 / 4.0
BEng in CSE
Kyushu Institute of TechnologyDepartment of Artificial Intelligence, School of Computer Science and Systems Engineering
Thesis: "Proposal of Representation-Conditional Latent Diffusion Models: Hematoma Classification in Brain CT Images"GPA: 3.7 / 4.0
High School Diploma
Kanonji Daiichi High School (Designated as a Super Science High School)Science and Mathematics Course
GPA: 4.9 / 5.0
Work Experience
Teaching Assistant (TA)
Lecturer: Tsuyoshi Okita
- Deep Learning Practicum (2026)
- Advanced Topics in Deep Learning I (2026)
- Deep Learning Practicum (2025)
- Deep Learning (2024)
Research Assistant (RA)
Supervisor: Tsuyoshi Okita
Conducted research and development on foundation models
OPTiM Corporation
Research Intern / Part-time Research Assistant
Conducted research and development on AI technologies for DX in agriculture
Social Experience
Student Volunteer (SV)
The 14th International Symposium on Applied Engineering and Sciences (SAES 2026) · Fukuoka, Japan
- Oral Session Chairperson
- Other support
Student Volunteer (SV)
- Registration desk
- Session support
- Excursion support
- Other support
Publications
- [1]
Weight Tying Causes Grokking in Conditional Flow Matching
Kosuke Ukita, Tsuyoshi Okita
The 14th International Symposium on Applied Engineering and Sciences (SAES 2026), No. p0092
November 2026 · Fukuoka, Japan - [2]
拡散モデルを用いた人間行動モーションデータからの写実的な映像生成
松石康希, 浮田嵩祐, 大北剛
情報処理学会論文誌 (IPSJ Journal), 67-10, 2026.
July 2026 · Japan - [3]
Efficient Inference for Flow Matching via Unified Path CFG
Kosuke Ukita, Tsuyoshi Okita
The 30th Pacific-Asia Conference on Knowledge Discovery and Data Mining (PAKDD 2026)
June 2026 · Hong Kong, China - [4]
High-Performance Self-Supervised Learning by Joint Training of Flow Matching
Kosuke Ukita, Tsuyoshi Okita
The 29th International Conference on Artificial Intelligence and Statistics (AISTATS 2026)
May 2026 · Tangier, Morocco - [5]
From Sensor Data to Sequence of Scenes: Diffusion Model-Based Visualization of Human Activity
Koki Matsuishi, Kosuke Ukita, Tsuyoshi Okita
IEEE Xplore, International Conference on Activity and Behavior Computing (ABC 2026), pp. 1-10, DOI: 10.1109/ABC68169.2026.11567210
March 2026 · Hokaido, Japan - [6]
Tsuyoshi Okita, Kosuke Ukita, Asahi Miyazaki, Daichi Kubota, Jukichi Ota, Naoki Kagiyama, Asahi Nishikawa, Daichi Nagayasu, Syunya Tomitaka, Daisuke Nozaki, Yuki Odo, Raku Yamashita, Xiaolong Ye, Huayu Gao, Kazuki Okahashi, Koki Matsuishi, Masaharu Kagiyama, Kodai Hirata, Haruki Kai, Lin Wang, Hristijan Gjoreski, Mathias Ciliberto, Paula Lago, Kazuya Murao, Daniel Roggen
Companion of the 2025 ACM International Joint Conference on Pervasive and Ubiquitous Computing (UbiComp '25), pp. 983–991
October 2025 · Espoo, Finland - [7]
Multimodal Foundation Model for Cross-Modal Retrieval and Activity Recognition Tasks
Koki Matsuishi, Kosuke Ukita, Tsuyoshi Okita
International Journal of Activity and Behavior Computing (IJABC), Care XDX Center, Kyushu Institute of Technology, Vol. 2025, No. 3, pp. 1-25
November 2025 · Kitakyushu, Japan - [8]
Energy-Efficient High-Performance Representation Learning for Sensor Data via Flow Matching
Kosuke Ukita, Tsuyoshi Okita
The 7th International Symposium on Neuromorphic AI Hardware, Paper No. P-19
October 2025 · Kitakyushu, Japan - [9]
Image Classification Using a Diffusion Model as a PreTraining Model
Kosuke Ukita, Xiaolong Ye, Tsuyoshi Okita
arXiv preprint arXiv:2505.06890, pp. 1-10
May 2025 - [10]
AURA-MFM: 詳細な行動理解のためのCLIPを用いたマルチモーダル基盤モデルの構築
松石康希, 浮田嵩祐, 大北剛
情報処理学会研究報告(IPSJ SIG Technical Report), Vol. 2025-UBI-85, No. 8
February 2025 · Osaka, Japan - [11]
Improving Human Activity Recognition through Diffusion Models
Kosuke Ukita, Tsuyoshi Okita
The 12nd International Symposium on Applied Engineering and Sciences (SAES 2024), No. p0052
November 2024 · Fukuoka, Japan - [12]
Ensemble Learning of Models for Human Activity Recognition
Asahi Miyazaki, Kosuke Ukita, Tsuyoshi Okita
The 12nd International Symposium on Applied Engineering and Sciences (SAES 2024), No. p0053
November 2024 · Fukuoka, Japan - [13]
Analysis of Human Activity Recognition by Diffusion Models
Kosuke Ukita, Tsuyoshi Okita
Companion of the 2024 ACM International Joint Conference on Pervasive and Ubiquitous Computing (UbiComp '24), pp. 458–463
October 2024 · Melbourne, Australia - [14]
浮田嵩祐, Ye Xiaolong, 大北剛
情報処理学会研究報告, 火の国情報シンポジウム 2024, No. A4-4, pp. 1-8
March 2024 · Kagoshima, Japan - [15]
Towards LLMs for Sensor Data: Multi-Task Self-Supervised Learning
Tsuyoshi Okita, Kosuke Ukita, Koki Matsuishi, Masaharu Kagiyama, Kodai Hirata, Asahi Miyazaki
Adjunct Proceedings of the 2023 ACM International Joint Conference on Pervasive and Ubiquitous Computing & the 2023 ACM International Symposium on Wearable Computing (UbiComp/ISWC '23 Adjunct), pp. 499–504
October 2023 · Cancun, Mexico
Grants
- –
Japan Science and Technology Agency (JST)
April 2026 - March 2029 · JPY ¥1,800,000 (+ ¥900,000) / year
- –
- –
Awards
- –
- –
- –
- –
Skills
Languages
Python · C++ · C · Java · JavaScript · TypeScript · Go · SQL · R
Frameworks & Tools
PyTorch · TensorFlow · Hugging Face · Jupyter · ABCI · Docker · Git
Other Skills
Deep Learning · Neural Networks · Model Training · Data Processing · API Development · Linux