Neural Network Verification
I study certified robustness for deep neural networks, with a particular emphasis on MaxPool-based architectures and tighter linear approximation methods.
Trustworthy AI · Verification · Security
Yuan Xiao
I am currently a Ph.D. student at the Software Institute, Nanjing University, under the supervision of Professor Zhenyu Chen and Associate Professor Chunrong Fang, with academic collaboration and guidance from Assistant Professor Shiqing Ma at the University of Massachusetts Amherst. I was also a joint Ph.D. student at the School of Computing, National University of Singapore, from November 2024 to November 2025, under the supervision of Professor Jinsong Dong, with collaboration and guidance from Professor Jun Sun. I received my bachelor's degree in Mathematics from Sun Yat-sen University in 2021, advised by Professor Xianping Guo.
Research
Current focus
Code Dataset Watermarking
I analyze how watermarks are embedded, detected, removed, and strengthened in code datasets and neural code models.
AI Security
I work on practical attack and defense problems around backdoors, unauthorized model usage, and trustworthy deployment of learning systems.
Education
Academic background
Ph.D., Nanjing University
Software Institute. Research on trustworthy AI, verification, and security.
Joint Ph.D., NUS
School of Computing. Collaboration with Professor Jinsong Dong and Professor Jun Sun.
B.S., Sun Yat-sen University
Mathematics. Advised by Professor Xianping Guo.
Publications
Selected papers across verification, watermarking, and security
† co-first authors. * corresponding authors. For Findings of ACL, the CCF list does not count it as a regular ranked conference, so it is marked as N/A.
Watermarking for Code Large Language Models and Datasets
DeCoMa: Detecting and Purifying Code Dataset Watermarks through Dual Channel Code Abstraction
International Symposium on Software Testing and Analysis, 2025 · Trondheim, Norway
Train in Vain: Functionality-Preserving Poisoning to Prevent Unauthorized Use of Code Datasets
Findings of the Association for Computational Linguistics, 2026
DuCodeMark: Dual-Purpose Code Dataset Watermarking via Style-Aware Watermark–Poison Design
Proceedings of the ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering, 2026
PuzzleMark: Implicit Jigsaw Learning for Robust Code Dataset Watermarking in Neural Code Completion Models
Proceedings of the ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering, 2026
Robustness Verification for Neural Network
Tightening Robustness Verification of MaxPool-based Neural Networks via Minimizing the Over-Approximation Zone
Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2025 · Tennessee, US
Towards General Robustness Verification of MaxPool-based Convolutional Neural Networks via Tightening Linear Approximation
Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2024 · Seattle, US
Backdoor Research for Neural Network
Talks
Recent presentations
Talks and posters
Highlights
Poster presentation at CVPR, plus an active publication pipeline across top AI and software engineering venues.
Service
External Review
Gallery
Snapshots from research and presentation
ISSTA 2025 Oral Presentation
A snapshot from my oral presentation at ISSTA 2025, highlighting active engagement with the software engineering research community.
ISSTA 2025 Attendance
A photo from my trip to ISSTA 2025.
CVPR 2024 Poster
A poster presentation snapshot from CVPR 2024, representing my work on robustness verification for neural networks.
Teaching
Teaching and academic activity
Teaching
Foundation of Data Science
Awards and projects
Awards
Projects
Contact
Open to postdoc and collaboration opportunities
If you are working on trustworthy AI, neural verification, or security for learning systems, feel free to reach out by email or check my full academic record in the CV page.