Doehyeon Lee | Seoul National University, Information Management Lab

About Me

I am a MS student in Information Management Lab at Seoul National University, with a strong interest in studying and exploring generation models in depth.

Research Interests

Multimodal AI Safety: Red-teaming multimodal models, Diagnosing safety failures, Building Safeguard
Human-Aligned Safety Evaluation: Reliable safety evaluation across modalities

News

[Oct. 2026] Two educational resources about AI Safety Direction and unified multimodal model are accepted to NeurIPS 2026, Education Track.
[Jan. 2026] One paper about Video Generation Model Jailbreak is accepted to ICLR 2026.
[May. 2025] One paper about VLM Safety Evaluation is accepted to ICML 2025.

Publications

ICML 2025

ELITE: Enhanced Language-Image Toxicity Evaluation for Safety

Wonjun Lee*, Doehyeon Lee*, Eugene Choi, Sangyoon Yu, Ashkan Yousefpour, Haon Park, Bumsub Ham, Suhyun Kim (*Equal Contribution)

Forty-second International Conference on Machine Learning, 2025 (ICML 2025)

PDF Code Project Page BibTex Poster Presentation
ICLR 2026

Jailbreaking on Text-to-Video Models via Scene Splitting Strategy

Wonjun Lee*, Haon Park*, Doehyeon Lee, Bumsub Ham, Suhyun Kim (*Equal Contribution)

Fourteenth International Conference on Learning Representations, 2026 (ICLR 2026)

PDF Code Project Page BibTex Poster Presentation

Conference Presentations

NeurIPS 2026 Edu

AI Safety Lives in a Direction: Reading, Breaking, and Repairing Refusal in Multimodal Models

Doehyeon Lee, Yoo-Min Jung, Jaeyong Ko

The Fortieth Annual Conference on Neural Information Processing Systems, Education Track, 2026 (NeurIPS 2026 Education Track)
NeurIPS 2026 Edu

Can a model think VISUALLY?

Jaeyong Ko, Eunseo Kim, Doehyeon Lee

The Fortieth Annual Conference on Neural Information Processing Systems, Education Track, 2026 (NeurIPS 2026 Education Track)

Korean Conference Presentations

이도현, 김이경, 성민재, 박종헌, “데이터 합성 기법을 활용한 음악 장르 분류 모델의 성능 향상 연구”, 대한산업공학회 춘계학술대회, 2025
성민재, 김이경, 이도현, 박종헌 “Genre Classification in Beat-aligned Monophonic Vocal Melody using Machine Learning”, 대한산업공학회 춘계학술대회, 2025
김이경, 성민재, 이도현, 박종헌, “LLM-Guided Music Tagging for Enhancing Semantic Consistency and Generalization in Tag Classification”, 대한산업공학회 춘계학술대회, 2025

Experience

Educations

M.S., Industrial Engineering (Information Management Lab), Seoul National University (In Progress)
B.S., Industrial Engineering, Seoul National University, 2025

Work Experience

Nanyang Technological University Singapore · Global Connect Fellowship (Feb 2026 – Mar 2026)
AIM Intelligence · Research Intern (Oct 2024 – Feb 2025)
NAVER Cloud · Intern, Hyperscale AI Planning (Jul 2024 – Sep 2024)
Seoul National University MARG (Music and Audio Research Group) · Student Research Intern (Jul 2023 – Aug 2023)

Academic Services

Reviewer for Trustworthy AI for Good (AI4GOOD) Workshop @NeurIPS 2026, @ICML 2026
Teaching Assistant, Industrial Application of Machine Learning, Seoul National University (Sep 2025 – Dec 2025)
Tutor, Basic Computing: First Adventures in Computing, Seoul National University (Mar 2025 – Dec 2025, Sep 2026 – Dec 2026)