Members

PhD Students

Master Students

Publication

1. Jaesin Ahn and Heechul Jung, "Mitigating Sexual Content Generation via Embedding Distortion in Text-conditioned Diffusion Models", NeurIPS 2025 (acceptance rate: 24.52%)

2. Ahn, Jaesin, and Heechul Jung. "Distorting Embedding Space for Safety: A Defense Mechanism for Adversarially Robust Diffusion Models." arXiv preprint arXiv:2501.18877 (2025).

3. Ahn, Jaesin, and Heechul Jung. "LLM Privacy Leakage Risk Analysis and the Effect of Privacy Pseudonymization." Proceedings of Symposium of the Korean Institute of communications and Information Sciences(2024).

Research Project(done)

1. "생성형AI 관련 프라이버시 리스크 경감기술 평가연구(Evaluation on Mitigation Technologies for Privacy Risks in Generative AI)", 개인정보보호위원회, May 2024 - Oct 2024

2. "소아 복부 멀티모달 및 합성데이터 (Pediatric Abdominal Multimodal and Synthetic Data)", NIA, May 2024 - Dec 2024

3. "2024년 데이터 문제해결은행 상황별 맞춤 지원 용역 비정형데이터 재현화 수행(Unstructured Data Synthesis/Reproduction for Problem-Solving Data Bank Project)", Aug 2024 - Oct 2024

4. "내시경 이미지 합성데이터(Endoscopic Image Synthetic Data)", NIA, July 2023 - Dec 2023

Patent

1. "부적절한 컨텐츠 생성을 유도하는 적대적 공격에 강건한 모델을 위한 임베딩 공간 왜곡 기법", 출원 준비중

Challenge

Model Repository

1. Korean Personal Data Anonymization (NER) Model

2. Measuring the Memorization of Korean Personal Data Based on Training Data Duplication