Research Theme
Trustworthy AI
- PhD Student in Data Science and Analytics at The Hong Kong University of Science and Technology (Guangzhou).
- Advised by Prof. Xinlei He (Wuhan University, Overseas High-Level Young Talent Program), with primary supervision from Prof. Weijia Heng and co-supervision from Prof. Yutao Yue in HKUST(GZ).
Research Interests
Trustworthy AI
AI Security & Privacy
AI-Generated Content Detection
AI For Safety
News
Paper Updates
- 🎉2026.01JALMBench was accepted at ICLR 2026.
- 🎉2025.126DAttack was accepted at AAAI 2026 as an Oral presentation.
- 🎉2025.09FC-Attack was accepted at EMNLP 2025 Findings.
- 🎉2025.09CHASM was accepted at NeurIPS 2025.
- 🎉2025.06Unsafe LLM-Based Search was accepted at USENIX Security 2025.
- 🎉2025.05AIGT on Social Media was accepted at ACL 2025, and MGT Generalization plus TH-Bench were accepted at KDD 2025.
- 🎉2025.03PEFTGuard was accepted at IEEE S&P 2025.
Awards and Milestones
- 🏆2025.12I received the 2025 DSA Excellent Research Award.
- 🏆2024.11AdSpectorX received the Best Paper Award at SENSYS-SocialMeta 2024.
- 🏆2024.06I received my firm PhD offer from HKUST(GZ).
Selected Papers
* denotes equal contribution. † denotes corresponding author.
IEEE S&P 2025 CCF-A
PEFTGuard: Detecting Backdoor Attacks Against Parameter-Efficient Fine-Tuning
ACL 2025 CCF-A
Are We in the AI-Generated Text World Already? Quantifying and Monitoring AIGT on Social Media
EMNLP 2025 Findings CCF-B
FC-Attack: Jailbreaking Large Vision-Language Models via Auto-Generated Flowcharts
KDD 2025 CCF-A
TH-Bench: Evaluating Evading Attacks via Humanizing AI Text on Machine-Generated Text Detectors
AAAI 2026 CCF-A · Oral
6DAttack: Backdoor Attacks in the 6DoF Pose Estimation
ICLR 2026CCF-A
JALMBench: Benchmarking Jailbreak Vulnerabilities in Audio Language Models
KDD 2025 CCF-A
On the Generalization and Adaptation Ability of Machine-Generated Text Detectors in Academic Writing
USENIX Security 2025 CCF-A
Unsafe LLM-Based Search: Quantitative Analysis and Mitigation of Safety Risks in AI Web Search
NeurIPS 2025 CCF-A
CHASM: Unveiling Covert Advertisements on Chinese Social Media
SENSYS-SocialMeta 2024 Best Paper
AdSpectorX: A Multimodal Expert Spector for Covert Advertising Detection on Chinese Social Media
Services
Conference PC / Reviewer
- The Web Conference 2025 Web4Good Track
- AAAI
- ACM MM
- ICML
- CVPR
- ACL
- EMNLP
- SaTML
- EuroS&P
- AsiaCCS
Journal Reviewer
- IEEE Transactions on Dependable and Secure Computing (TDSC)
- IEEE Transactions on Information Forensics and Security (TIFS)
- ACM Transactions on Privacy and Security (TOPS)
- International Journal of Human-Computer Interaction (IJHCI)
Honors and Awards
- 2025, DSA Excellent Research Award
- Kaggle Competitions Expert (Vincent Sirius)
- 2020.04, MCM/ICM Meritorious Winner
- 2019 / 2020 / 2021, Third-class Scholarship of BUPT
- 2019 / 2020 / 2021, Excellent Student Leader of BUPT
Education
- 2024.08-present, PhD in Data Science and Analytics, The Hong Kong University of Science and Technology (Guangzhou)
- 2022.08-2023.10, MSc in Computer Science, City University of Hong Kong
- 2018.09-2022.07, BSc in Computer Science and Technology, Beijing University of Posts and Telecommunications
Experience
- Research Assistant, 2023.06-2024.05, Centre for Artificial Intelligence and Robotics (CAIR), Hong Kong Institute of Science & Innovation, Chinese Academy of Sciences (HKISI-CAS). Worked on surgical LLMs and image segmentation. Supervisor: Dr. Jinlin Wu.
- Project Participant, 2022.09-2023.08, City University of Hong Kong. Worked on financial machine translation. Supervisor: Prof. Linqi Song.