I am currently a Ph.D. candidate in the School of Data Science (SDS), the Chinese University of Hong Kong, Shenzhen (CUHKSZ) under the supervision of Professor Hongyuan Zha and Professor Baoyuan Wu. Previously, I received my Bachelor’s degree (rank No.1, first-class honor) in Electronic Information Engineering from the School of Science and Engineering (SSE), CUHKSZ.

Research

My research interests mainly lie in AI Security and Fairness, Computer Vision and Optimization, but also include Kernel Methods, Reinforcement Learning, and the application of Machine Learning in economics/marketing.

I am currently seeking opportunities for postdoctoral/research positions in the field of machine learning.

News

2024

• [2024-09-26] Two papers get accepted by NeurIPS 2024.
• [2024-08-25] I released some useful tools for academic writing and rebuttal at here.
• [2024-05-23] I leave SCLBD, grateful for the invaluable experiences gained.
• [2024-01-16] One paper gets accepted by ICLR 2024, in close collaboration with Zihao Zhu and Mingda Zhang.
• [2024-01-13] I receive the Best Poster Award in The 3rd Doctoral & Postdoctoral Academic Forum held by Shenzhen Research Institute of Big Data.

2023

• [2023-12-27] I receive the Guo Tai Jun An Scholarship to recognize my outstanding research achievements from 2020-2023.
• [2023-12-15] We release Defenses in Adversarial Machine Learning: A Survey, a comprehensive survey for defense methods in adversarial machine learning. The draft is available at this link.
• [2023-12-15] I receive the 2023 Duan Yong Ping Travel Award to recognize my outstanding research achievements in academic year 2022-2023.
• [2023-10-03] I was invited as a guest speaker for the tutorial Backdoor Learning: Recent Advances and Future Trends in ICCV 2023. Slides are available at this link.
• [2023-09-22] Two papers get accepted by NeurIPS 2023, in close collaboration with Mingli Zhu and Mingda Zhang.
• [2023-07-18] One paper gets accepted by ICCV 2023, in close collaboration with Mingli Zhu.
• [2023-04-21] BackdoorBench is updated to the 2nd version with more methods and analysis tools involved. The code and data are available at this link.
• [2023-01-20] One paper gets accepted by AISTATS 2023, in close collaboration with Jiayin Liu.

Fundings and projects

• Daoyuan Youth Fund Project - Class I (道远I类青年基金项目)

Awards

• Best Poster Award in The 3rd Doctoral & Postdoctoral Academic Forum
• 2023 Guo Tai Jun An Scholarship
• 2023 Duan Yong Ping Travel Award
• AIRS Talent of Ph.D. Research Program of Academic Year of 2020-2021

Patents

• Shaokui Wei, Baoyuan Wu, Mingda Zhang, Hongyuan Zha. Method and system for eliminating shared adversarial samples in backdoor defense (一种后门防御中消除共享对抗样本的方法及后门防御系统). China. Patent No. CN117390622A. Jan. 12, 2024.
• Mingli Zhu, Shaokui Wei, Baoyuan Wu. Method and system for backdoor defense by purifying toxic features through neural polarizers (通过神经偏振器净化有毒特征的后门防御方法及系统). China. Patent No. CN116629319A. Aug. 22, 2023.
• Baoyuan Wu, Mingli Zhu, Shaokui Wei, Li Shen, Yanbo Fan. Backdoor defense method, terminal device, and computer-readable storage medium. China (后门防御方法、终端设备及计算机可读存储介质). Patent No. CN116578974A. Aug. 11, 2023.

Selected Papers

(*Indicates equal contribution, #Indicates corresponding author)

• Mitigating Backdoor Attack by Injecting Proactive Defensive Backdoor.
  Authors: Shaokui Wei, Hongyuan Zha, Baoyuan Wu.
  Publication: NeurIPS 2024.
  Links: Paper | Code | Website.

• Shared Adversarial Unlearning: Backdoor Mitigation by Unlearning Shared Adversarial Examples.
  Authors: Shaokui Wei, Mingda Zhang, Hongyuan Zha, Baoyuan Wu.
  Publication: NeurIPS 2023.
  Links: Paper | Code | Website.

• Neural Polarizer: A Lightweight and Effective Backdoor Defense via Purifying Poisoned Features.
  Authors: Mingli Zhu * , Shaokui Wei * , Hongyuan Zha, Baoyuan Wu.
  Publication: NeurIPS 2023.
  Links: Paper | Code | Website.

• Mean Parity Fair Regression in RKHS.
  Authors: Shaokui Wei, Jiayin Liu, Bing Li, Hongyuan Zha.
  Publication: AISTATS 2023.
  Links: Paper | Code | Website.

• Backdoorbench: A comprehensive benchmark of backdoor learning.
  Authors: Baoyuan Wu, Hongrui Chen, Mingda Zhang, Zihao Zhu, Shaokui Wei, Danni Yuan, Chao Shen.
  Publication: NeurIPS 2022 Dataset and Benchmark Track.
  Links: Paper | Code | Website.

• Unveiling and Mitigating Backdoor Vulnerabilities based on Unlearning Weight Changes and Backdoor Activeness.
  Authors: Weilin Lin, Li Liu, Shaokui Wei, Jianze Li, Hui Xiong
  Publication: NeurIPS 2024.
  Links: Paper | Code | Website.

• VDC: Versatile Data Cleanser for Detecting Dirty Samples via Visual-Linguistic Inconsistency.
  Authors: Zihao Zhu, Mingda Zhang, Shaokui Wei, Bingzhe Wu, Baoyuan Wu.
  Publication: ICLR 2024.
  Links: Paper | Code | Website.

• Enhancing Fine-Tuning Based Backdoor Defense with Sharpness-Aware Minimization.
  Authors: Mingli Zhu, Shaokui Wei, Li Shen, Yanbo Fan, Baoyuan Wu.
  Publication: ICCV 2023.
  Links: Paper | Code | Website.