Xiaoyun Xu

Xiaoyun Xu

Assistant Professor at Zhongguancun Academy

About Me

I’m currently an Assistant Professor at Zhongguancun Academy. Before joining Zhongguancun Academy, I obtained my PhD from the Digital Security group at Radboud University, where I was fortunate to be advised by Stjepan Picek. I also spent eight months as a postdoctoral researcher at Radboud University.

Before that, I earned my MSc in Advanced Computing from the University of Bristol and a BEng in Software Engineering from the University of Electronic Science and Technology of China. I also spent two years as a Research Assistant at ISCAS before starting my PhD.

Research Interests:

My research interests lie primarily in adversarial machine learning and designing robust and general defenses against risks in machine learning technologies.

News

  • 2026.09: One paper accepted to S&P 2027.
  • 2026.06: One paper accepted to UIST 2026.
  • 2026.06: TPC member @ USENIX Security 2027.
  • 2025.11: MIMIR in NDSS 2026.
  • 2025.09: TPC member @ CCS 2026.
  • 2025.05: One paper accepted to CCS 2025.
  • 2025.02: MIMIR achieved TOP 1 on RobustBench ImageNet Leaderboard
  • 2024.09: BAN in NeurIPS 2024

Explore the Town

Publications

NDSS 2026
sym

MIMIR: Masked Image Modeling for Mutual Information-based Adversarial Robustness

Xiaoyun Xu, Shujian Yu, Zhuoran Liu, Stjepan Picek

Network and Distributed System Security (NDSS) Symposium, 2026

Code arXiv

AE AvailableAE FunctionalAE Reproduced
CCS 2025
sym

Towards Backdoor Stealthiness in Model Parameter Space

Xiaoyun Xu, Zhuoran Liu, Stefanos Koffas, Stjepan Picek

ACM Conference on Computer and Communications Security (CCS), 2025

Code arXiv

AE AvailableAE FunctionalAE Reproduced
NeurIPS 2024
sym

BAN: Detecting Backdoors Activated by Adversarial Neuron Noise

Xiaoyun Xu, Zhuoran Liu, Stefanos Koffas, Shujian Yu, Stjepan Picek

Advances in Neural Information Processing Systems (NeurIPS), 2024

Code arXiv

DSN 2024
sym

Universal Soldier: Using universal adversarial perturbations for detecting backdoor attacks

Xiaoyun Xu, Oguzhan Ersoy, Behrad Tajalli, Stjepan Picek

IEEE/IFIP International Conference on Dependable Systems and Networks Workshops (DSN-W), 2024

arXiv

CCS 2023
sym

Poster: Boosting Adversarial Robustness by Adversarial Pre-training

Xiaoyun Xu, Stjepan Picek

ACM Conference on Computer and Communications Security (CCS), 2023

DSN 2023
sym

IB-RAR: Information Bottleneck as Regularizer for Adversarial Robustness

Xiaoyun Xu, Guilherme Perin, Stjepan Picek

IEEE/IFIP International Conference on Dependable Systems and Networks Workshops (DSN-W), 2023

Code arXiv

CCS@PPLMP 2020
sym

Information leakage by model weights on federated learning

Xiaoyun Xu, Jingzheng Wu, Mutian Yang, Tianyue Luo, Xu Duan, Weiheng Li, Yanjun Wu, Bin Wu

In Proceedings of the 2020 workshop on privacy-preserving machine learning in practice, CCS workshop PPLMP, 2020

Educations

  • 2022 - 2025, PhD student, Radboud University
  • 2017 - 2018, Master, University of Bristol
  • 2013 - 2017, Undergraduate, University of Electronic Science and Technology of China (UESTC)

Services

  • Reviewer/PC: USENIX Security, CCS, BMVC, ICLR, NeurIPS, SaTML, TIFS

  • External Reviewer: IEEE SP, NDSS

  • Session Chair: SaTML

Miscellaneous

  • Outside of research, I enjoy seeing the world, exploring diverse cultures and local cuisines, and taking in landscapes.
  • The Legend of Zelda, Sid Meier’s Civilization VI, The Binding of Isaac, Don’t Starve.