RIKEN Center for Advanced Intelligence Project Uncertainty Quantification Team

Team Director: Futoshi Futami (Ph.D.)

Japanese Page

Research Summary

As machine learning is increasingly applied in high-stakes domains, it is critical not only to ensure accuracy but also to quantify predictive uncertainty. Our team develops theoretical frameworks and algorithms to evaluate and control uncertainty, using tools from statistical learning theory, information theory, and Bayesian statistics. We focus on calibration of predicted probabilities, epistemic uncertainty, and latent variable models. By deepening the mathematical foundations of these topics, we aim to advance the development of reliable machine learning systems with rigorous uncertainty quantification.

Main Research Fields

Informatics

Related Research Fields

Engineering
Mathematical & Physical Sciences
Intelligent informatics-related
Theory of informatics-related
Theory of informatics-related

Keywords

Machine learning
Bayesian inference
Uncertainty evaluation

Selected Publications

1. F. Futami.
"Epistemic Uncertainty and Excess Risk in Variational Inference."
To be appeared in Artificial Intelligence and Statistics, 2025. (In press).
2. F. Futami & M. Fujisawa.
"Information-theoretic Generalization Analysis for Expected Calibration Error."
Advances in Neural Information Processing Systems, 37, 84246--84297, 2024.
3. F. Futami & T. Iwata.
"Information-theoretic Analysis of Bayesian Test Data Sensitivity."
Proceedings of 27th International Conference on Artificial Intelligence and Statistics, 238, 1099-1107, 2024.
4. F. Futami & M. Fujisawa.
"Time-Independent Information-Theoretic Generalization Bounds for SGLD",
Advances in Neural Information Processing Systems, 36, 8173-8185, 2023.

Lab Members

Principal investigator

Futoshi Futami: Team Director

Contact Information

English School of Engineering Science, Graduate School of Engineering Science
1-3, Machikaneyama, Toyonaka, Osaka, Japan.