YUGA IGUCHI / 井口 優雅
Hello! Welcome to my website!
Overview:
Currently, I am a Senior Research Associate at the School of Mathematical Sciences, Lancaster University, as part of the Prob_AI Hub, working with Paul Fearnhead.
PhD in Statistical Science at UCL. Supervisors: Alexandros Beskos and Samuel Livingstone.
Before my PhD, I worked as a research assistant at Hitotsubashi University under the supervision of Toshihiro Yamada, and subsequently as a quantitative analyst at MUFG Bank, Ltd. for two years.
I am an Associate Editor for Statistics and Computing (2025-).
Research: I work at the intersection of Statistics, Probability and Numerical Analysis. My interest lies in diffusion processes and their related topics, including
Statistical inference: parameter estimation, filtering problems, asymptotic analysis of estimators;
Numerical analysis: development of numerical schemes and error analysis;
Computational statistics: sampling from a target distribution via Langevin diffusions;
Probabilistic AI: mathematical underpinning of generative diffusion models.
I have been developing numerical schemes and statistical estimators to address practical scenarios where the standard approximation, e.g., the Euler-Maruyama scheme, can be inaccurate or break down. Recently, I have also been working on mathematical and computational foundations of generative AI.
Contact: y[dot]iguchi[at]lancaster[dot]ac[dot]uk or yuga[dot]iguchi[dot]21[at]alumni[dot]ucl[dot]ac[dot]uk
News:
2026 August
I have been invited to Dual Trimester Program: Geometric Statistics: theory, application, and computation @ Hausdorff Research Institute for Mathematics. I will talk about my research on hypoelliptic diffusions and some related open problems!
Our new preprint `Diffusion Models for High-Dimensional Clustered Data: Intrinsic-Dimension Adaptivity via Bayesian Classification' is now available.
2026 July
Our paper `A closed-form transition density expansion for elliptic and hypoelliptic SDEs' has appeared in Bernoulli.
Our new preprint `Pathwise skew-symmetric discretisation for SDEs with superlinear drift' is now available.
I gave a talk at the ISBA satellite meeting at the Institute of Statistical Mathematics, Tachikawa!