Shot with Sony a7iii and FE 1.2/50 GM by Pengrun Huang
1/4000s, f/1.2, ISO 100

I’m Yiwei (yee-way) Fu (foo), a first year CS PhD student at University of Illinois Urbana-Champaign (UIUC). Previously, I had a master’s degree in machine learning at Carnegie Mellon University. Before that, I graduated from University of Michigan majoring in Honors Mathematics and Computer Science. I also worked at W. W. Grainger as an applied machine learning scientist.

My research interest is in the privacy, safety, and security of machine learning, especially their connections with each other. I’m broadly interested in differential privacy, safety, and alignment of generative models with their information theoretic implications, and efficient and secure machine learning systems.

More of me here. Non-academic stuff here.

Email: yiwei_fuscz AT outlook.com / fyw AT umich.edu

[CV] [Bluesky] [LinkedIn]


Research

Unlearning or Obfuscating? Jogging the Memory of Unlearned LLMs via Benign Relearning
Shengyuan Hu, Yiwei Fu, Zhiwei Steven Wu, Virginia Smith
The Thirteenth International Conference on Learning Representations, 2025
[PDF]

BLUR: A Benchmark for LLM Unlearning Robust to Forget-Retain Overlap
Shengyuan Hu, Neil Kale, Pratiksha Thaker, Yiwei Fu, Zhiwei Steven Wu, Virginia Smith
Preprint, [PDF]

Enhancing One-run Privacy Auditing with Quantile Regression-Based Membership Inference
Terrance Liu, Matteo Boglioni, Yiwei Fu, Shengyuan Hu, Pratiksha Thaker, Zhiwei Steven Wu
Preprint, [PDF]


Notes and Blogs