About Me
Hi~ I am Zixiong Wang (王子雄) 🐻
- I am currently a 2nd year PhD student jointly trained by Huazhong University of Science and Technology (HUST) and PKU-Wuhan Institute for Artificial Intelligence (Whai-PKU).
- I am working at the School of Electronic Information and Communications (EIC), supervised by Prof. Gaoyang Liu.
- Before that, I have received my Bachelor’s degree from Central China Normal University (CCNU) in 2022.
- Previously, I work as an NLP researcher student, focusing on data privacy and security issues (e.g., copyright, unlearning, and data privacy) of LLMs and relevant applications (e.g., long-context modeling, Reasoning, and Agent System).
What’s New 🔥
[8/2026] Our Paper “Black-Box Membership Inference Attacks against Contrastive Learning via Aggressive Data Augmentations” got accepted in IEEE TDSC🎉
Sincere appreciations to Prof. Gaoyang Liu~ 💝
[5/2026] Our Paper “Fingerprinting Pre-trained Encoders under Arbitary Downstream Fine-tuning via Adversarial Shifting” got accepted in ICML 2026🎉
Sincere appreciations to Dr. Tianlong Xu~ 💝
[11/2025] Our paper “Breaking the Boundary Barrier: Robust Model Fingerprinting via Unlearnable Examples in Model-Parameter Space” got accepted in ACM SIGKDD 2026🎉
Sincere appreciations to Dr. Tianlong Xu~ 💝
[10/2025] My teammates and me participated in the Challenge Cup 2025 Games (2025“挑战杯-揭榜挂帅”擂台赛)! We won the Grand Prize (国赛特等奖) in AI Special Track @Shanghai (ranked 2nd in the SH-04 track). Congratulations🎉
[5/2025] Our paper “Decoupling Memories, Muting Neurons: Towards Practical Machine Unlearning for Large Language Models” got accepted in ACL 2025 Findings🎉
Sincere appreciations to Dr. Lishuai Hou~ 💝
[3/2025] Our paper “Prototype Surgery: Tailoring Neural Prototypes via Soft Labels for Efficient Machine Unlearning” got accepted in ACM CCS 2025🎉
Sincere congratulations to Dr. Xijie Wang~ 💝
[11/2024] Our preprint paper “Membership Inference Attack against Long-Context Large Language Models” has been released to Arxiv🎉
My Research Areas 🔍
- Privacy risks and Data Security Protection of Large Language Models
- Membership Inference Attacks on ML models
- Training Data Extraction or Recovering on LLMs
- LLM/VLM Unlearning
- Model/Data Watermark & Fingerprint of LLMs
- Security of Large Reasoning Models and LLM Agents
More Info 📧
- Feel free to contact me if you feel my works inspiring or interesting, my email address is zixwang1015@hust.edu.cn.
- My working area is at the East#17 building of Huazhong University of Science and Technology, C309 office.
- The Github page of our HUST EIC Trustworthy AI Lab is https://github.com/SPHelixLab.
- Hope to make more friends!