Sanghyeok Chu

Ph.D. Candidate · Seoul National University

I am a Ph.D. candidate in Electrical Engineering and Computer Science at Seoul National University, advised by Prof. Bohyung Han. My research focuses on developing scalable and efficient foundation models, with a particular interest in vision-language models, efficient architectures, and long-context video understanding. I am currently seeking research internship opportunities for Summer/Fall 2026.

Email Scholar LinkedIn

News

2026.02 Two papers accepted to CVPR 2026.
2025.06 Started internship at LG AI Research.
2025.05 One paper accepted to ICML 2025.

Experience

LG AI Research, Physical Intelligence Lab (formerly Vision Lab)

Research Scientist Intern

Research on integrating the Mixture-of-Experts architecture into vision encoders for next-generation foundation models (2025.06 – 2025.11)
Research on linear-time memory architectures for streaming VideoLLMs (2025.12 – 2026.05)

2025.06 – 2026.05

Seoul, Korea

Publications

(* denotes equal contribution)

Under Review, 2026
Interference-Aware DeltaNet for Streaming Data

Sanghyeok Chu, Pyunghwan Ahn, Gwangmo Song, Seung Hwan Kim, Honglak Lee, and Bohyung Han

Paper Project
arXiv, 2026
Pri4R: Learning World Dynamics for Vision-Language-Action Models with Privileged 4D Representation

Jisoo Kim*, Jungbin Cho*, Sanghyeok Chu, Ananya Bal, Jinhyung Kim, Gunhee Lee, Sihaeng Lee, Seung Hwan Kim, Bohyung Han, Hyunmin Lee, Laszlo A. Jeni, Seungryong Kim

Paper Project
CVPR 2026
Enhancing Mixture-of-Experts Specialization via Cluster-Aware Upcycling

Sanghyeok Chu, Pyunghwan Ahn, Gwangmo Song, Seung Hwan Kim, Honglak Lee, and Bohyung Han

Paper Project Code
CVPR 2026
Beyond the Ground Truth: Enhanced Supervision for Image Restoration

Donghun Ryou*, Inju Ha*, Sanghyeok Chu, and Bohyung Han

Paper Project Code
ICML 2025
Fine-Grained Captioning of Long Videos through Scene Graph Consolidation

Sanghyeok Chu, Seonguk Seo, and Bohyung Han

Paper Project Code
CVPR 2022
Pooling Revisited: Your Receptive Field is Suboptimal

Donghwan Jang, Sanghyeok Chu, Joonhyuk Kim, and Bohyung Han

Paper
MVA 2022
Beyond Homography: Nonparametric Image Alignment via Graph Convolutional Networks

Mijeong Kim, Sanghyeok Chu, and Bohyung Han

Paper
NeurIPS 2021
Learning Debiased and Disentangled Representations for Semantic Segmentation

Sanghyeok Chu, Dongwan Kim, and Bohyung Han

Paper Code

Awards

3rd Place, OmniLabel Challenge 2023 (Track C)

1st OmniLabel Workshop on Infinite Label Spaces for Semantic Understanding via Natural Language, at CVPR 2023

Education

Seoul National University

M.S. & Ph.D. integrated course in Electrical Engineering and Computer Science

Advisor: Prof. Bohyung Han

2019.09 – Present

Seoul, Korea

Yonsei University

B.S. in Electrical and Electronic Engineering

2012.03 – 2019.09

Seoul, Korea