Nonparametric Learning
& Optimization Lab
Welcome to the Nonparametric Learning & Optimization Lab.
I am Jae-Hwan Jhong, leading this lab. Since 2020 I have been with the Department of Information Statistics at Chungbuk National University, where I have served as an Associate Professor since 2024. I received my Ph.D. in Statistics from Korea University in 2019, with a dissertation titled A Study on Penalized B-splines for Function Estimation, under the supervision of Professor Ja-Yong Koo. Prior to joining Chungbuk National University, I served as a Research Professor at the Department of Statistics, Korea University.
Our lab develops nonparametric statistical methodologies centered on function estimation, with a strong emphasis on the penalization and optimization techniques that make these methods both computationally effective and theoretically well-founded. We work at the interface of methodology, theory, and application, and we welcome collaborations across statistics, data science, and the domain sciences.
This website serves as a hub for introducing our research, teaching, and lab members. Materials such as publications, preprints, and research notes are available upon request. If you are interested in our work, would like to collaborate, or are considering joining the lab, please feel free to reach out.
Research Interests
Nonparametric Function Estimation
Much of our work centers on flexible function estimation that does not rely on rigid parametric assumptions. We have developed a suite of methods based on regression splines — including B-splines, I-splines, and polygram regression — that adapt naturally to complex structures in data such as quantile surfaces, spherical geometries, single-index models, and monotone or shape-constrained regression.
Penalization & Optimization
A central tool in our work is penalization. We design and study penalties tailored to specific estimation problems — total variation, lasso, group lasso, adaptive lasso, nuclear norm, and non-negative garrote, among others — and develop the convex and non-convex optimization algorithms needed to compute them efficiently. The resulting estimators are simultaneously flexible, interpretable, and tractable.
Statistical Learning & Applications
We investigate the theoretical properties — consistency, convergence rates, adaptivity — of the estimators we develop, and apply them to problems in biostatistics, environmental science, and beyond. Recent collaborations include modeling the impact of air pollution on lung function, oceanographic and ecological curve fitting, and spatiotemporal models for environmental exposure data.
Join Us
We welcome motivated students interested in nonparametric statistics, statistical learning, and optimization to join our lab. If you are considering graduate study at the Department of Information Statistics, or would like to discuss collaborations, please contact Jae-Hwan Jhong at jjh25@cbnu.ac.kr.