Junsuk Choe Multimodal AI Lab (MAIL)

Associate Professor

Junsuk Choe

Multimodal AI for Intelligent Systems

Department of Computer Science and Engineering, Sogang University

Office: AS-913 · Phone: +82-2-705-8495

Email Scholar CV Calendar LinkedIn

I am an associate professor in the Department of Computer Science and Engineering at Sogang University. Before joining Sogang University in Fall 2021, I spent two wonderful years as a Research Intern and then a Research Scientist at NAVER AI Lab. I received my Ph.D. from Yonsei University in 2020 under the supervision of Prof. Hyunjung Shim, and my B.S. from Yonsei University in 2013.

Prospective Students

Recruiting one graduate student, starting March 2027

We welcome inquiries from students interested in:

  • Agentic AI
  • Robotics
  • Efficient LLM inference
Get in touch by email

Updates

News

  1. Minyoung visits Johns Hopkins University as a visiting researcher, hosted by Prof. Jaemin Cho.
  2. Jimin visits the University of British Columbia as a visiting researcher, hosted by Prof. Cristina Conati.
  3. Three papers accepted to ICML 2026 workshops.
  4. One paper accepted to ICPR 2026.
  5. One paper accepted to the Findings track of CVPR 2026.
  6. Two papers accepted to ICLR 2026.
  7. We have 9 papers: 1 AAAI, 1 ICML, 1 COLM, 1 NeurIPS, 4 Pattern Recognition Letters, and 1 Neurocomputing.
  8. Promoted to Associate Professor at Sogang University. Thanks to all my students, colleagues, and collaborators for their support.
  9. Yeji joined NAVER AI Lab as a research intern.
  10. Granted the 생성AI 선도인재양성사업 as Co-PI, led by NC AI and funded by IITP, and the 독자 AI 파운데이션 모델 프로젝트 as Co-PI, led by Upstage and funded by NIPA/NIA/IITP.
  11. Dongjun visited the University of Tübingen as a visiting researcher, hosted by Prof. Seong Joon Oh.
  12. Minyoung, Yeji, and Yejin visited the University of Tübingen as visiting researchers, hosted by Prof. Seong Joon Oh.
  13. Hyo Seo visited Michigan State University as a visiting researcher, hosted by Prof. Sijia Liu.
  14. Granted the Global Humanities and Social Convergence Research Program (1.8B KRW) as Co-PI from NRF Korea, led by Prof. Dasaem Jeong.
  15. Dongjun participated in the IITP AI Intensive Education Program at Carnegie Mellon University.
  16. Granted the Outstanding Young Scientist Grant (913M KRW) as PI from NRF Korea, with Prof. Buru Chang as Co-PI.
  17. Minyoung and Hyo Seo visited the University of Tübingen as visiting researchers, hosted by Prof. Seong Joon Oh and Prof. Zeynep Akata.
  18. Joined the Department of Computer Science and Engineering at Sogang University as Assistant Professor.
  19. Defended Ph.D. thesis and joined NAVER AI Lab as Research Scientist.

Focus

Research Interests

My research centers on large multimodal foundation models — vision-language, video-language, and increasingly agentic models that plan and act in digital and physical environments — and aims to make them sustainable to manage and reliable to deploy without costly retraining. I pursue this goal along three complementary directions: unlearning and continual learning, faithful and efficient inference, and agentic and embodied AI.

Unlearning and Continual Learning

I develop methods for changing what a large model already knows — removing facts it should forget, absorbing skills it lacks — with a growing focus on structures that keep knowledge localized and composable rather than entangled.

Faithful and Efficient Inference

I work on keeping multimodal outputs grounded in the intended evidence — addressing hallucination and cross-modal imbalance — while designing inference-time compression that scales with input complexity instead of breaking under it.

Agentic and Embodied AI

As foundation models move from passive perception to autonomous action, I study how they can plan, use tools, and act reliably — as digital agents navigating multi-step tasks or embodied vision-language-action models in the physical world.

Lab

People

Current Members

6
  • Dongjun Hwang
    PhD 24.03-MS 22.03-24.02Intern 21.09-22.02
    AI Seoul Tech Scholarship (PhD)NRF Scholarship (PhD)Smilegate DHE Scholarship (MS)
    MS AI, BS CSE, Sogang Univ.Visiting Researcher, Univ. of Tübingen (25.05-26.07)CMU IITP AI Education Program (24.09-25.02)Research: Continual learning for VLA
  • Minyoung Lee
    PhD 25.03-MS 23.03-25.02Intern 22.03-23.02
    AI Seoul Tech Scholarship (PhD)Smilegate DHE Scholarship (MS)
    MS AI, BS CSE, Sogang Univ.Visiting Researcher, Johns Hopkins Univ. (26.08-26.11)Visiting Researcher, Univ. of Tübingen (25.01-25.02, 23.07-23.08)Research: 3D motion for VLA
  • Yeji Park
    Integrated MS-PhD 23.03-Intern 22.10-23.02
    Hyundai Motor Chung Mong-Koo Scholarship
    BE Econ, BS Math, Sogang Univ.Research Intern, NAVER AI Lab (25.09-26.02)Visiting Researcher, Univ. of Tübingen (25.01-25.02)Research: Multi-instance KV compression
  • Jimin Hong
    MS 26.03-Intern 25.03-26.02
    BS CSE, Sogang Univ.Visiting Researcher, Univ. of British Columbia (26.08-26.11)Research: MLLM KV compression
  • Jihong Park
    Intern 26.01-
    BS ME, Sogang Univ., exp. 27.02Research: Travel agents
  • Seungjun Ha
    MS 26.09-Intern 26.02-26.08
    Sogang Albatross Fellowship
    BS CSE, Sogang Univ.Research: Egocentric vision

Alumni

5
  • Elena Kuular
    MS 25.10-26.08Intern 25.04-25.09
    GKS Scholarship
    MS CSE, Sogang Univ.BS Econ, Moscow State Univ.
  • Yejin Kim
    MS 24.03-26.02Intern 23.02-24.02
    Smilegate DHE ScholarshipSogang Albatross Fellowship
    MS AI, BS CSE, Sogang Univ.Visiting Researcher, Univ. of Tübingen (25.01-25.02)Now: KAIST AI, Research Intern
  • Hyo Seo Kim
    MS 22.09-25.08
    MS CSE, BS Econ & CSE, Sogang Univ.Visiting Researcher, Michigan State Univ. (24.10-25.04)Research Intern, NAVER AI Lab (24.07-24.10)Visiting Researcher, Univ. of Tübingen (23.07-23.08)Now: Illinois Tech, PhD student
  • Doyeol Baek
    MS 22.09-24.08Intern 22.01-22.08
    Smilegate DHE Scholarship
    MS AI, BS CSE, Sogang Univ.Now: Samsung Electronics, SWE
  • Junyong Kang
    Intern 21.12-23.03
    Valedictorian (College of Engineering)
    BS CSE, Sogang Univ.Now: KAIST AI, MS student

Resources

Equipment

Our on-premise inventory keeps growing, and we actively supplement it with cloud GPU services such as RunPod for large-scale or bursty compute needs.

  • GPU
    96GB: NVIDIA RTX Pro 6000 ×7 48GB: A6000 ×8, RTX Pro 5000 ×2 24GB: A5000 ×16, RTX 4090 ×5, RTX 3090 ×2
  • Robot
    Franka Research 3 ×1

Service

Academic Activities

Area Chair

  • COLM 2025-26
  • ICLR 2026-27
  • NeurIPS 2026

Reviewer

  • Computer Vision: CVPR, ICCV, ECCV, WACV, ACCV
  • Machine Learning / AI: ICML, ICLR, NeurIPS, CoLLAs, AAAI, AISTATS, IJCAI
  • Journals: IEEE PAMI, IJCV, IEEE TIP, IEEE TCSVT

Organizer

  • Korean Conference on Computer Vision, 2025-26
  • 대한전자공학회 영상처리연구회 겨울학교, 2024-25
  • 대한전자공학회 영상이해 및 영상처리연구회 합동 여름학교, 2023-24-25

Output

Publications

J/C/W/P denote journal, conference, workshop, and preprint. * co-first, ^ co-corresponding.

P5

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

Minyoung Lee, Jimin Hong, Junsuk Choe

Under review, 2026.

P4

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

Jimin Hong, Minyoung Lee, Junsuk Choe

Under review, 2026.

P3

Mitigating Cross-Image Information Leakage in Multi-Image Understanding with Large Vision-Language Models

Yeji Park, Minyoung Lee, Sanghyuk Chun, Junsuk Choe

Under review, 2026.

P2

Knowledge Vector Weakening: Efficient Training-free Unlearning for Large Vision-Language Models

Yejin Kim, Dongjun Hwang, Sungmin Cha, Junsuk Choe

Under review, 2026.

P1

ALISE: Alignment-aware Data Selection for Unlearning in Contrastive Vision-Language Models

Dongjun Hwang, Yejin Kim, Beomyun Kwon, Junsuk Choe

Under review, 2026. Extended version of the ICML 2026 MemFM workshop paper.

C20

RARR: Real-time Attention-Driven Rain Removal with Hierarchical Scale-aware Efficient Network

Seungho Eum, Ihjoon Cho, Jeonghyeon Kim, Junsuk Choe^, Unsang Park^

ICPR 2026.