Hanjin Lee 이한진
Integrated M.S./Ph.D. in Computer Science · AISLab, Yonsei University
I work on understanding, controlling, and securing modern generative models — especially diffusion-based systems — from a principled, mathematical perspective. My goal is AI that is trustworthy, controllable, and secure by design.
What I work on
Bridging theoretical rigor and practical system reliability — so AI behaves safely even under distribution shift or adversarial conditions.
Generative Model Safety
Securing and controlling diffusion-based systems with distribution-preserving mechanisms that keep outputs reliable.
Adversarial Robustness
Studying adversarial examples, patches, and perturbations — how they break vision systems, and how to defend against them.
AI for Security
Applying ML to malware classification, anomaly detection, and defending models against extraction & inversion attacks.
Selected & full publications
Reverse-chronological. Bold marks my name. Peer-reviewed journals, conferences, and preprints.
Recent updates
Background
Education
Integrated M.S./Ph.D., Computer Science (AI)
B.S. in Computer Engineering
Experience
Graduate Researcher — AI security & generative models
Undergraduate Researcher
Recognition
10+ best-paper and presentation awards across Korean AI & signal-processing societies.
Let's talk research 🙌
Open to collaboration on generative-model safety and AI security.