Minyoung Lee

I am a Ph.D. student in Computer Science and Engineering at Sogang University, advised by Junsuk Choe. My research centers on multimodal foundation models, combining text, vision, audio, and action to build systems that are interpretable, efficient, and reliable in practice. My recent interests:

Interpretability of multimodal models and how they integrate information across modalities
Robustness and generalization under real-world distribution shifts
Bridging vision-language understanding with action and control for robotic systems
Efficient learning methods that improve models without additional data or compute

News

[Jan. 2026] 1 paper accepted at ICLR 2026.
[Sept. 2025] 1 paper accepted at NeurIPS 2025.
[Jan. 2025] 1 paper accepted in Pattern Recognition Letters (SJR Q1; IF: 3.9).
[Jan. 2025] Served as a visiting scholar at the Tübingen AI Center (Host: Prof. Dr. Seong Joon Oh).
[Jul. 2023] Served as a visiting scholar at the Tübingen AI Center (Hosts: Prof. Dr. Seong Joon Oh and Prof. Dr. Zeynep Akata).

Education

Ph.D. in Computer Science and Engineering, Sogang University, Seoul (Mar. 2025 – ) Advisor: Junsuk Choe
M.S. in Artificial Intelligence, Sogang University, Seoul (Mar. 2023 – Feb. 2025) Advisor: Junsuk Choe
B.S. in Computer Science and Engineering, Sogang University, Seoul (Mar. 2019 – Feb. 2023) Honors: magna cum laude

Work Experience

Visiting Researcher, Tübingen AI Center (Jan. 2025 – Feb. 2025) Host: Prof. Dr. Seong Joon Oh
Visiting Researcher, Tübingen AI Center (Jul. 2023 – Aug. 2023) Host: Prof. Dr. Seong Joon Oh and Prof. Dr. Zeynep Akata).

Open-Source Contributions

lmms-eval — Contributed the Mantis multi-image benchmark.

Teaching Assistant

Basic Machine Learning (Spring 2023)
Hacking and Information Security (Fall 2023)

Awards and Honors

Digital Human & Entertainment Scholarship, Smilegate, Korea (2023 – 2024)
AI Seoul Tech Graduate Scholarship, Seoul Future Talent Foundation, Korea (2026)
Jang Jik-hyun Honorary Professor Scholarship, Sogang University, Korea (2022)

Publications

Inserting Random Tokens at the Tail: A Simple Training-Free Input Perturbation for LVLMs

Jimin Hong, Minyoung Lee, Junsuk Choe

Preprint.

Enhancing Cross-Modal Interactions in Audio-Visual LLMs via Training-Free Attention Injection

Minyoung Lee, Jimin Hong, Junsuk Choe

Preprint.

Mitigating Cross-Image Information Leakage in LVLMs for Multi-Image Tasks

Yeji Park, Minyoung Lee, Sanghyuk Chun, Junsuk Choe

Preprint.

PDF Code

Enhancing Multi-Image Understanding Through Delimiter Token Scaling

Minyoung Lee, Yeji Park, Dongjun Hwang, Yejin Kim, Seong Joon Oh, Junsuk Choe

International Conference on Learning Representations (ICLR), 2026.

PDF Code

OVS Meets Continual Learning: Towards Sustainable Open-Vocabulary Segmentation

Dongjun Hwang, Yejin Kim, Minyoung Lee, Seong Joon Oh, Junsuk Choe

Conference on Neural Information Processing Systems (NeurIPS), 2025.

PDF Code

Fog-Free Training for Foggy Scene Understanding

Minyoung Lee, Kyungwoo Song, Junsuk Choe

Pattern Recognition Letters, 2025. (SJR Q1; IF 3.9)

PDF ACCV 2024 Workshop on Women in Computer Vision (WiCV)

Services

Conference Reviewers

Conference on Neural Information Processing Systems (NeurIPS) 2026