I am a M.S. student at CVLab in Yonsei University, advised by Prof. Bumsub Ham, and also affiliated with KDST in Kyung Hee University (formerly at KIST), co-advised by Prof. Suhyun Kim.

I am broadly interested in building trustworthy and safe AI that can operate reliably under real-world conditions.

My recent research interests include the following topics:

• Trustworthy and Safe AI: Jailbreak, Safety Evaluation, Benchmark, Human Alignment
• Improving the Capabilities of AI: Transformer-based Understanding, LLM & VLM Reasoning

🔥 News

• 2025.09:  New preprints are out: SceneSplit, ReSafe
• 2025.05:  One paper is accepted to ICML 2025
• 2025.02:  New preprint is out: ELITE
• 2024.12:  One paper is accepted to AAAI 2025

📝 Publications

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

  Wonjun Lee*, Haon Park*, Doehyeon Lee, Bumsub Ham, Suhyun Kim (* Equal Contribution)
  Preprint, 2025
  [Paper]

• ReSafe: Enhancing Safety of Text-to-Image Diffusion via Post-Hoc Image Back Translation
  

  Hyunsoo Kim, Wonjun Lee, Donghyun Kim^†, Suhyun Kim^†
  Preprint, 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)
  International Conference on Machine Learning (ICML) 2025
  [Paper] / [Project Page]

• Maximizing the Position Embedding for Vision Transformers with Global Average Pooling
  

  Wonjun Lee, Bumsub Ham, Suhyun Kim
  Association for the Advancement of Artificial Intelligence (AAAI) 2025
  [Paper] / [Project Page]

📖 Educations

• 2024.03 - Present, M.S. in Electrical and Electronic Engineering, Yonsei University.
• 2018.03 - 2024.02, B.S. in Software Convergence, Kyung Hee University.