Design Deep Learning
like Mathematical Models

Deep learning excels at learning diverse mappings, but its success relies heavily on the mathematical framework in which it operates. By incorporating principles from differential geometry and Bayesian statistics, deep learning becomes a flexible mathematical model that we can tailor to our specific needs.

Welcome to SAIL.

1 Mar. 2025
We invite international students and research interns to join our group.

23 Jan. 2025
Mr. Khosrovian has had his paper accepted to ICLR2025.

10 Dec. 2025
Prof. Matsubara has had his paper accepted to AAAI2025 as an oral presentation.

26 Nov. 2024
We are currently seeking a Research Fellow in AI for Science.

About Us Members

Research

Scientific Machine Learning

As part of "AI for Science," we advance data-driven modeling and accelerate computational simulations.

Geometric Deep Learning

We leverage the geometric structures inherent in data to enable faithful generation and robust recognition across diverse domains.

Bayesian Deep Learning

We design deep learning frameworks grounded in Bayesian modeling for reliable small-scale data analysis and decision-making.

Miscellaneous

We explore constraints and supports to deep learning architectures and optimizations.

ResearchPublications

News

Jan. 2025
Mr. Razmik Arman Khosrovian has had his paper, "Poisson-Dirac Neural Networks for Modeling Coupled Dynamical Systems across Domains," co-authored with Prof. Yoshimura, Waseda University, and Prof. Yaguchi, Kobe University, accepted at The Thirteenth International Conference on Learning Representations (ICLR2025). PoDiNNs uncover coupling patterns and component-wise properties in coupled dynamical systems, offering precise forecasts and interpretability.
Dec. 2024
Prof. Takashi Matsubara has had his paper, "Number Theoretic Accelerated Learning of Physics-Informed Neural Networks," co-authored with Prof. Yaguchi, Kobe University, accepted at The Thirty-Ninth AAAI Conference on Artificial Intelligence (AAAI2025). The number-theoretic method selects an optimal set of collocation points, accelerating PINNs' training.
Oct. 2024
Mr. Razmik Arman Khosrovian and Mr. Yosuke Nishimoto have had their papers accepted at the NeurIPS 2024 Workshops on Machine Learning and the Physical Sciences, and Compositional Learning, respectively.
Jun. 2024
Prof. Takashi Matsubara will give an invited talk at CAI 2024 Workshop on Scientific Machine Learning and Its Industrial Applications (SMLIA2024) at Singapore.
Apr. 2024
Prof. Takashi Matsubara have been appointed as a professor at the Faculty of Information Science and Technology, Hokkaido University.
Mar. 2024
Prof. Takashi Matsubara will give a tutorial talk at International Conference on Scientific Computing and Machine Learning (SCML2024), held at Kyoto.
Feb. 2024
Mr. Kota Sueyoshi has had his paper, "Predicated Diffusion: Predicate Logic-Based Attention Guidance for Text-to-Image Diffusion Models," accepted at IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR2024). (Update: Selected as a highlight.)
News