Naoya Takeishi is a researcher at the University of Tokyo. His interests include machine learning for scientific problems (currently on hybrid modeling and simulation-based inference) and the data-driven analysis of dynamical systems.
Please also see his research group’s website.
Contact
ntake[at]g.ecc.u-tokyo.ac.jp
The inboxes of old addresses (…@ailab.t.u-tokyo.ac.jp, …@hesge.ch, …@riken.jp) are no longer monitored.
Recent activities
See publications for papers and other presentations.
| 2026-08-26 | Lecture on simulation-based inference @ AI for Science Summer School, RIKEN iTHEMS |
|---|---|
| 2026-03-11 | Lecture on dynamic mode decomposition @ Spring School on Computational Physics 2026 |
| 2026-03-03 | Invited talk on machine learning utilizing physical information @ The 13th SICE Multi-Symposium on Control Systems [slides] |
| 2025-03-01 | Invited talk on hybrid modeling @ the Workshop on Functional Inference and Machine Intelligence [slides] |
| 2024-12-14 | I will be a panelist @ the workshop on Machine Learning and the Physical Sciences, NeurIPS 2024 |
| 2024-10-22 | Invited talk on ML and scientific models @ The 2024 Fall Meeting of the Seismological Society of Japan [slides] |
| 2024-09-23 | I joined the Editorial Board of the journal Machine Learning: Science and Technology (MLST) |
| 2024-07-15 | Invited talk on neural nets and Koopman operator learning @ the workshop on Koopman Operators in Robotics, RSS 2024 [slides] |
| 2024-05-29 | Tutorial talk on ML and scientific models @ The 38th Annual Conference of the Japanese Society for Artificial Intelligence [slides] |
| 2023-08-03 | Invited talk on ML and scientific models @ The 46th Annual Meeting of the Japan Neuroscience Society |
| 2023-07-28 | Co-organized the workshop on Synergy of Scientific and ML Modeling @ ICML 2023 |
Research interests
ML for Science: Machine learning and scientific models
keywords: hybrid modeling, simulation-based inference, physics-informed machine learning
Exploring how scientific mathematical models / simulators relate to data-driven machine learning models. What does each type of model represent, and how is it obtained? How can the two be combined? Can a data-driven model be “understood” like a scientific model?
Hybrid modeling (grey-box modeling) combines scientific and machine learning models. Examples include hybrid generative models and methods for properly learning hybrid models.
Machine learning approaches to forward problems (e.g., solving differential equations) and inverse problems (e.g., inferring simulation parameters). Recent work focuses on simulation-based inference (SBI), including reliable neural SBI and multifidelity SBI for cosmology.
Data-driven dynamical systems
keywords: Koopman operator, dynamic mode decomposition
Data-driven analysis of dynamical systems, especially dynamic mode decomposition (DMD) and its underlying theory based on the Koopman operator. Methods developed include Bayesian DMD, DMD for random dynamics, DMD with neural net observables, nonnegative DMD, time-varying DMD, and discriminant DMD for labeled time-series.