About me

Hi! 👋 I am a research assistant in CITI at Academia Sinica, working with Prof. Jun-Cheng Chen. Previously, I served as a research assistant at the Institute of Information Science, Academia Sinica, under the supervision of Prof. Chun-Shien Lu and co-advised by Prof. Chia-Mu Yu during this time.

I obtained two M.Sc. degrees: one in Computer Science from National Chengchi University, Taiwan (advised by Prof. Raylin Tso), and another in Electrical Engineering & Computer Science from Kanazawa University, Japan (advised by Prof. Masahiro Mambo). I received my B.Sc. degree in Information Management from Chang Gung University, Taiwan.

I am looking for a Ph.D. position. I will be thrilled if you are interested in my profile. Feel free to contact me by email!

🔬 Research Interest

My primary research interests focus on developing trustworthy and reliable AI systems to advance technology and address social challenges. I aim to ensure robustness, privacy, fairness, and interpretability of AI when applied across various domains.

  • Generative AI: diffusion models, multimodal models, large language models.
  • Trustworthy AI: adversarial robustness, security, privacy-preserving techniques.
  • Others: copyright protection, robust generalization, safety alignment.

📢 News

  • 2025-07-12: Our paper is accepted to ECAI 2025: "BadVim: Unveiling Backdoor Threats in Visual State Space Model".
  • 2024-09-01: Our paper is accepted to WACV 2025: "Defending Against Repetitive Backdoor Attacks on Semi-supervised Learning through Lens of Rate-Distortion-Perception Trade-off".
  • 2024-04-12: Our paper is accepted to ICME 2024 [Oral]: "On The Higher Moment Disparity of Backdoor Attacks".

📝 Professional Service

  • Conference Reviewer
    • IEEE ICME (2024, 2025)

For more info


Feel free to reach out at chengyi.lee.1224 [AT] gmail.com for potential collaborations or discussions.