I am Xiaoyi Chen, a final-year Ph.D. candidate in Peking University advised by Prof. Zhonghai Wu. I was a visiting student at CISPA - Helmholtz Center for Information Security during October 2019 - October 2020, advised by Prof. Michael Backes and Dr. Yang Zhang. Previously, I got my bachelor degree in Chongqing University in 2016 and I graduated from Zhengzhou Foreign Language school, Henan, China in 2012.

My research interests widely spread in the intersection areas of AI and security, including security and privacy of machine learning, and the application of machine learning techniques in security related domains. Specifically, recent days I focus on the backdoor attacks and defenses against DNN models.

Research interests

  • Machine Learning for Cyber Security
  • Backdoor Attacks and Defenses
  • Adversarial Robustness

Publications

  • BadNL: Backdoor Attacks against NLP models with Semantic-preserving Improvements
    Xiaoyi Chen, Ahmed Salem, Dingfan Chen, Michael Backes, Shiqing Ma, Qingni Shen, Zhonghai Wu, Yang Zhang
    2021 Annual Computer Security Applications Conference (ACSAC ’21)
    [pdf][slides][trace]

  • BadNL: Backdoor Attacks against NLP models
    Xiaoyi Chen*, Ahmed Salem*, Michael Backes, Shiqing Ma, Yang Zhang
    Workshop on Adversarial ML at ICML 2021 (ICML-W ’21)
    [pdf][poster]

Experience

Research Intern

Microsoft Research Asia

Work on a research project about optimization of Microsoft Azure Application Gateway.

Mentor: Dr. Yongqiang Xiong (Networking Research Group)

August 2017 - February 2018

Visiting Student

CISPA - Helmholtz Center for Information Security

Work on a research project about backdoor attacks against NLP models..

Advisor: Prof. Michael Backes and Dr. Yang Zhang

October 2019 - October 2020

Education

Peking University

Doctor of Philosophy
Information Security, Computer Science
2016 - 2022 (expected)

Chongqing University

Bachelor
Information Security, 94.5/100, 1/284
2012 - 2016