Wenbo Zhou is an Associate Professor at the University of Science and Technology of China, specializing in AI security, particularly in the areas of Deepfake generation and detection. He holds a B.S. from Nanjing University of Aeronautics and Astronautics (2014) and a Ph.D. from the University of Science and Technology of China (2019). He is an IEEE member and an influential researcher in AI security. Zhou has won multiple prestigious awards, including the “Distinguished Artifact Award” at ACM CCS. He was part of the team that won second place in the world in the Deepfake Detection Challenge (DFDC), earning a prize of 300,000 US dollars. His development of DeepFaceLab, a globally recognized Deepfake tool, has cemented his place as a leader in the field of AI security. Zhou has also published widely in high-impact journals and conferences, contributing significantly to advancements in AI and cybersecurity.
Profile
Education
Wenbo Zhou received his B.S. degree from Nanjing University of Aeronautics and Astronautics, Nanjing, China, in 2014. Following his undergraduate studies, he pursued his Ph.D. at the University of Science and Technology of China, Hefei, China, and completed his doctoral degree in 2019. His research during his Ph.D. focused on AI security, laying the foundation for his future work in Deepfake detection and adversarial machine learning. Zhou’s academic journey has been marked by a blend of rigorous coursework and groundbreaking research. His Ph.D. work, combined with hands-on experience in AI security tools like DeepFaceLab, set him apart as a leader in the field. Throughout his education, Zhou demonstrated a commitment to advancing technology for practical applications, as evidenced by his multiple patents and innovations in AI security.
Experience
Wenbo Zhou is currently an Associate Professor at the University of Science and Technology of China (USTC). He has led over 10 research projects funded by the Natural Science Foundation of China, with a total funding exceeding ¥20 million. His extensive experience in AI security includes significant contributions to the detection and generation of Deepfakes, with his tools like DeepFaceLab gaining global recognition. Zhou has also been a visiting scholar at Microsoft Research, where he further refined his research on AI and cybersecurity. His work on various patents, such as those in Deepfake detection, shows his ability to bridge theoretical research with practical solutions. Zhou’s expertise extends to peer-reviewed publications in top-tier journals like IEEE Transactions on Information Forensics & Security and Pattern Recognition. His multidisciplinary approach and collaborations with both academia and industry have placed him at the forefront of AI security research.
Research Focus
Wenbo Zhou’s research focuses on AI security, with particular expertise in Deepfake generation and detection, adversarial examples, and steganography. His work addresses critical issues in digital forensics, such as the authentication of media and the detection of manipulated content. Zhou has made significant strides in Deepfake detection, contributing to the global conversation about digital disinformation and cybersecurity. His development of DeepFaceLab, one of the most influential Deepfake tools worldwide, has revolutionized the field of face-swapping and manipulation detection. In addition to Deepfakes, Zhou explores adversarial machine learning, aiming to defend AI systems against vulnerabilities exploited by malicious actors. His research also touches on areas like watermarking and the development of robust image processing techniques to combat the misuse of AI in creating counterfeit media. Zhou’s work not only advances theoretical AI security but also provides practical solutions for combating emerging threats in the digital world.
Publication Top Notes
- Multi-attentional deepfake detection 🧠
- Spatial-phase shallow learning: Rethinking face forgery detection in frequency domain 📡
- Dup-net: Denoiser and upsampler network for 3D adversarial point clouds defense 🖼️
- Model watermarking for image processing networks 💧
- Hairclip: Design your hair by text and reference image 💇♂️
- Finfer: Frame inference-based deepfake detection for high-visual-quality videos 🎥
- A new rule for cost reassignment in adaptive steganography 💻
- {X-Adv}: Physical adversarial object attacks against X-ray prohibited item detection 📦
- Initiative defense against facial manipulation 🧑⚖️