Juhyeong Kim — Researcher

01About

I am an M.S./Ph.D. student in the Optimization and Statistical Inference (OSI) Lab at KAIST AI (Kim Jaechul Graduate School of AI), advised by Prof. Se-Young Yun. I received my B.S. in Mathematical Science with a double major in Computer Science at KAIST. My research focuses on generative models — diffusion models, one-step generative modeling, and test-time alignment — as well as the mathematical foundations of bandit algorithms. I am also interested in robotics and diffusion policy.

Generative Models
Diffusion Models
Bandit Algorithms
Robotics

02Publications

^* Equal contribution · ^† Corresponding author

Comparison of reverse drift, forward drift, and DualDrift

PUBLICATION 01

DualDrift: Forward and Reverse Drifting Fields for One-Step Generative Modeling

Hojung Jung^*, Juhyeong Kim^*, Jaehyun Kwak, Boryeong Cho, Junhyeok Yang, Youngrok Park, Sangmin Bae, Se-Young Yun^†

NeurIPS 2026 (under review) · ICML 2026 Workshop (accepted)

A one-step generative modeling framework that jointly leverages forward and reverse drifting fields to produce high-quality samples without iterative generation.

UniSAFE safety risk comparison across tasks and models

PUBLICATION 02

UniSAFE: A Comprehensive Benchmark for Safety Evaluation of Unified Multimodal Models

Segyu Lee^*, Boryeong Cho^*, Hojung Jung^*, Seokhyun An, Juhyeong Kim, Jaehyun Kwak, Yongjin Yang, Sangwon Jang, Youngrok Park, Wonjun Chang, Se-Young Yun^†

Preprint, 2026

Paper Code & Data

A system-level safety benchmark for unified multimodal models spanning seven input/output settings and 6,802 curated instances.