RIKEN Center for Advanced Intelligence Project Dynamical Systems Learning Team

Team Director: Yoshinobu Kawahara (D.Eng.)

Japanese Page

Research Summary

Our team focuses on the dynamical structures underlying data-generating mechanisms, developing machine learning theories and algorithms for their prediction and control. At the same time, by leveraging the dynamical-system perspective of information processing in machine learning, we establish novel machine learning principles and develop models and algorithms based on these principles.

Research Subjects:

Machine learning based on dynamics in data generation mechanisms
Machine learning for analysis, prediction and control of dynamical systems
Application of developed methods to scientific and engineering data

Main Research Fields

Informatics

Related Research Fields

Engineering
Mathematical & Physical Sciences
Mathematical informatics
Statistical science
Intelligent informatics

Keywords

Machine Learning
Dynamical Systems
Nonlinear Dynamics
Physics-Informed Machine Learning

Selected Publications

1. Ke, N., Tanaka, R., and Kawahara, Y.:
"Learning Stochastic Nonlinear Dynamics with Embedded Latent Transfer Operators"
Proceedings of the 28th International Conference on Artificial Intelligence and Statistics, PMLR 258, pp.4861-4869 (2025).
2. Fujii, K., Takeuchi, K., Kuribayashi, A., Takeishi, N., Kawahara, Y., and Takeda, K.:
"Estimating Counterfactual Treatment Outcomes Over Time in Complex Multiagent Scenarios"
IEEE Transactions on Neural Networks and Learning Systems, Vol.36, No.2, pp.2103-2117 (2025).
3. Sakata, I. and Kawahara, Y.:
"Enhancing Spectral Analysis in Nonlinear Dynamics with Pseudoeigenfunctions from Continuous Spectra"
Scientific Reports, Vol.14, Article Number 19276 (2024).
4. Weissenbacher, M., Agarwal, R., and Kawahara, Y.:
"SiT: Symmetry-Invariant Transformers for Generalisation in Reinforcement Learning"
Proceedings of the 41st International Conference on Machine Learning, pp.52695-52719 (2024).
5. Ohnishi, M., Ishikawa, I., Lowrey, K., Ikeda, M., Kakade, S., and Kawahara, Y.:
"Koopman Spectrum Nonlinear Regulators and Efficient On-line Learning"
Transactions on Machine Learning Research (2024).
6. Konishi, T. and Kawahara, Y.:
"Stable invariant models via Koopman spectra"
Neural Networks 165, pp.393-405 (2023).
7. Iwata, T. and Kawahara, Y.:
"Neural dynamic mode decomposition for end-to-end modeling of nonlinear dynamics"
Journal of Computational Dynamics 10(2), pp.268-280 (2023).
8. Weissenbacher, M., Sinha, S., Garg, A., and Kawahara, Y.:
"Koopman Q-learning: Offline reinforcement learning via symmetries of dynamics"
Proceedings of the 39th International Conference on Machine Learning, pp.23645-23667 (2022).
9. Takeishi, N., and Kawahara, Y.:
"Discriminant Dynamic Mode Decomposition for Labeled Spatio-Temporal Data Collections "
SIAM Journal on Applied Dynamical Systems, Vol.21, No.2, pp.1030-1058 (2022).
10. Hashimoto, Y., Ishikawa, I., Ikeda, M., Komura, F., Katsura, T., and Kawahara, Y.:
"Reproducing kernel Hilbert C*-modules and kernel mean embeddings"
Journal of Machine Learning Research, Vol.22, No.267, pp.1-56 (2021).

Lab Members

Principal investigator

Yoshinobu Kawahara: Team Director

Core members

Pongpisit Thanasutives: Special Postdoctoral Researcher
Itsushi Sakata: Postdoctoral Researcher
Satoshi Noguchi: Postdoctoral Researcher
Yoshiteru Nishimura: Technical Staff I
Keisuke Fujii: Visiting Scientist
Yuka Hashimoto: Visiting Scientist
Takuya Konishi: Visiting Scientist
Masahiro Fujisawa: Visiting Scientist
Masanobu Horie: Visiting Scientist
Jingjing Bai: Research Part-time Worker I
Ryogo Tanaka: Research Part-time Worker I
Yuto Inui: Research Part-time Worker II
Yuta Miyauchi: Research Part-time Worker II

Contact Information

Graduate School of Information Science A Bldg.,
Suita Campus, Osaka University,
1-5 Yamadaoka, Suita-shi, Osaka
565-0871 Japan
Email: yoshinobu.kawahara@riken.jp