Hello, I'm

白浩 Hao Bai

Ph.D. Candidate @ Wuhan University
Focusing on cutting-edge research in AI Security, Machine Learning Privacy Protection, and Data Security. Committed to advancing technological innovation and contributing to both academia and industry.
Federated Learning Differential Privacy Membership Inference Secure Computation
View Research Publications Contact Me
白浩
3 Papers
10+ Citations

About Me

0
Published Papers
0
Citations
+
0
Research Projects
+
0
Years of Research

I am a researcher focused on AI security and machine learning privacy protection, currently a Ph.D. candidate at Wuhan University.

My research interests mainly focus on cutting-edge technologies including Homomorphic Encryption, Secure Multi-party Computation, Differential Privacy, Membership Inference Attacks, and Backdoor Attacks.

Currently, I have participated in multiple important research projects and published several high-quality academic papers in top-tier venues. I am committed to combining theoretical research with practical applications, promoting innovative applications of AI security and privacy protection technologies in various fields.

In addition to academic research, I actively participate in academic exchanges and collaborations, serving as a reviewer for international conferences and journals, contributing to the advancement of the field.

Technical Skills

Python
PyTorch
Federated Learning
Cryptography

Education

2024 - Present
In Progress
Ph.D. in Engineering
Wuhan University

Research Focus: Machine Learning Privacy Protection & AI Security

Federated Learning Differential Privacy Secure Computation
2022 - 2024
Completed
M.Eng. in Engineering
Wuhan University

Research Focus: Machine Learning Privacy Protection & AI Security

Privacy-preserving ML Neural Network Security
2018 - 2022
Completed
B.Eng. in Engineering
Wuhan University

Achievement: Top 1% in Major, Outstanding Graduate

Top 1% Outstanding Graduate

Research

Federated Learning
ACE-pFL: Personalized Federated Learning Framework

We propose a novel accurate and efficient personalized federated learning (pFL) framework based on Knowledge Distillation, called ACE-pFL.

  • EMD-based client clustering for non-IID data
  • Dynamic distillation temperature adjustment
  • Triple distillation strategy leveraging global and local knowledge
  • Achieves balance between accuracy, training time, and communication
Personalized FL Knowledge Distillation Non-IID Data
IEEE ToN (CCF-A)
Privacy-preserving ML
Privacy-preserving CNN Inference System

An efficient privacy-preserving convolutional neural network (CNN) inference scheme that exploits different computational characteristics of linear and non-linear layers.

  • Matrix split computing protocol design
  • Parameterized quadratic polynomial approximation for ReLU
  • Secret sharing based lightweight cryptographic primitives
  • 2-15x speedup with only ~2% accuracy loss
Privacy-preserving CNN Secret Sharing
Journal of Software (CCF-T1)
Differential Privacy
GetFed: Differentially Private FL with GAN

GetFed: An accurate and differentially private FL framework with GAN-based Data Generation on non-IID Data.

  • Integrated DP in GAN training and federated aggregation
  • Dynamic noise reduction as virtual sample quality improves
  • Adaptive DP-based secure aggregation algorithm
  • 6-47% accuracy improvement, 50% training time reduction
Federated Learning GAN Differential Privacy
IEEE TDSC (CCF-A)

Publications

01
CCF-A 2025
ACE-pFL: Accurate, Efficient Personalized Federated Learning with Knowledge Distillation
Kun He, Hao Bai, Yuqing Li, Jing Chen, Ruiying Du
IEEE Transactions on Networking
PDF Code Cite
02
CCF-T1 2025
Efficient Privacy-preserving Inference based on Secret Sharing for Convolutional Neural Networks (基于秘密分享的高效隐私保护卷积神经网络预测)
Hao Bai, Kun He, Jing Chen, Chenbin Zhao, Ruiying Du
Journal of Software (软件学报)
PDF Cite
03
CCF-A 2025
GetFed: Accurate, Differentially Private Federated Learning with GAN-based Data Generation
Hao Bai, Kun He, Yuqing Li, Jing Chen, Haowei Li, Zhongmou Liu, Xuanang Yang, Ruiying Du
IEEE Transactions on Dependable and Secure Computing
PDF Code Cite

Get In Touch

Feel free to reach out for collaborations, research discussions, or any questions!

Email
baihao@whu.edu.cn
Phone
+86 187-4823-1316
Address
Xinjia Building, National Cybersecurity Center
Dongxihu District, Wuhan, Hubei, China
GitHub
github.com/zxc1139135