Haiyun He
(Pronunciation: /ˈhaɪyuːŋ ˈhɜː/, Pronouns: she/her/hers)
Assistant Professor, Internet of Things Thrust, HKUST(GZ)
(Taken at Cornell Slope)
Hi! I am currently an Assistant Professor in the Internet of Things (IoT) Thrust at the Hong Kong University of Science and Technology (Guangzhou). I am also a Cross-Campus Faculty Affiliate at HKUST. I obtained my PhD degree in Electrical and Computer Engineering from National University of Singapore in Sep. 2022, advised by Prof. Vincent Y. F. Tan. Before joining HKUST(GZ), I was a postdoc in the Center for Applied Mathematics at Cornell University, working with Prof. Ziv Goldfeld from ECE and Prof. Christina Lee Yu from ORIE. I was honored to be funded by the Center for Applied Mathematics postdoctoral fellowship.
My recent research interests lie in the intersection of information theory and machine learning, including but not limited to:
- Machine learning and statistical learning theory
- Hypothesis testing
- Inference and estimation
- Trustworthy AI
If you are interested in collaborating with me, please do not hesitate to drop me an email (at the bottom) or drop by my office W3-604 !
- Theoretical research: For students interested in the mathematical foundations of machine learning, drawing on information theory, probability, and statistics. A strong mathematical background and an interest in developing rigorous proofs are especially welcome.
- Experimental research: For students interested in developing and evaluating methods for trustworthy machine learning and large language models. Strong programming skills and an interest in designing careful experiments and analyzing results are especially welcome.
[Year/Term]-[PhD/MPhil/RA Application]-[Your Name],
e.g., 27Spring-PhD Application-Haiyun He.
News
| 08/21/2026 |
 Congrats to my former RA Baixin Li on his first diffusion language model (DLM) watermarking paper accepted to EMNLP 2026 Findings, titled with “SAC-Copula: Quality-Preserving Watermarking for Diffusion Language Models via Smooth Correlated Gumbel Fields”. We study the problem of watermarking DLMs, which requires mechanisms compatible with iterative parallel unmasking rather than autoregressive decoding. We propose SAC-Copula, a quality-preserving watermarking method for DLMs based on smooth, locally correlated Gumbel perturbation fields constructed via a Gaussian copula.
|
|---|---|
| 07/31/2026 | Interested in keeping up with the latest GenAI watermarking papers? Check out the daily arXiv tracker: Watermarking Arxiv Daily. It is maintained by Zhenxin Ai. |
| 05/13/2026 |
Congrats to my incoming PhD student Zhenxin Ai on his first watermarking paper accepted to ICML 2026, titled with “PASA: A Principled Embedding-Space Watermarking Approach for LLM-Generated Text under Semantic-Invariant Attacks”. He also made an excellent project page. Come to chat with him in Seoul!
|