Jungbin Jun

I am an undergraduate student in Statistics at Inha University and an undergraduate researcher in the Theoretical Statistics Lab under the supervision of Prof. Ilsang Ohn.

My research interests are centered on developing reliable learning methods that quantify uncertainty and adapt to changing environments. My work on online conformal inference and variational Bayes has led me to explore related problems in PAC-Bayesian learning, reinforcement learning, and bandits.

Email  /  CV  /  Github  /  Blog

Profile photo

Education
Inha University
B.S. in Statistics
Mar 2022 - Sep 2026 (Expected)
  • GPA: 4.42 / 4.50 (Major: 4.47 / 4.50)
  • Early Graduation (3 years)
  • Rank 1/41 in the department

Publications
Online Conformal Inference with Retrospective Adjustment for Faster Adaptation to Distribution Shift
Jungbin Jun, Ilsang Ohn
arXiv preprint, 2025
paper / code

RetroAdj is an online conformal method that dynamically updates \(\alpha_t\) and retrospectively adjusts the prediction interval \(\hat C_t(\alpha_t)\) over time. It adapts quickly under distribution shift while remaining efficient and long-run valid.

Early-stopped Aggregation: Adaptive Inference with Computational Efficiency
Ilsang Ohn, Shitao Fan, Jungbin Jun, Lizhen Lin
arXiv preprint, 2026
paper

ESA is a variational Bayes aggregation method that constructs model-specific posteriors via a variational free energy and aggregates only promising models, avoiding unnecessary exploration of overly large models.

Reading Notes

Machine Learning

Statistics and Mathematics

Website template adapted from Jon Barron.