Mingyi Zhou

Mingyi Zhou was a PhD student at Monash University. His research project was ‘Reliable AI Deployment on Mobile Apps’.

His PhD focused on using SE techniques such as program analysis to enhance the reliability of deployed machine learning (ML) systems, especially for smart mobile apps. This includes assessing the risks associated with ML systems, protecting their intellectual property, and minimising their attacking surface. Additionally, he is exploring automated ways to optimise machine learning programs on mobile devices to improve inference efficiency.

Selected publications

• Mingyi Zhou, Xiang Gao, Xiao Chen, Chunyang Chen, John Grundy, and Li Li. "DynaMO: Protecting Mobile DL Models through Coupling Obfuscated DL Operators" IEEE/ACM International Conference on Automated Software Engineering (ASE’24).
• Mingyi Zhou, Xiang Gao, Pei Liu, John Grundy, Xiao Chen, Chunyang Chen, and Li Li. "ModelObfuscator: Obfuscating Model Information to Protect Deployed ML-based Systems" ACM SIGSOFT International Symposium on Software Testing and Analysis (ISSTA’24).
• Mingyi Zhou, Xiang Gao, Jing Wu, Kui Liu, Hailong Sun, and Li Li. "Investigating White-Box Attacks for On-Device Models" IEEE/ACM International Conference on Software Engineering (ICSE’24).
• Mingyi Zhou, Xiang Gao, Jing Wu, John Grundy, Chunyang Chen, Xiao Chen, and Li Li. "ModelObfuscator: Obfuscating Model Information to Protect Deployed ML-based Systems" ACM SIGSOFT International Symposium on Software Testing and Analysis (ISSTA’23).
• Mingyi Zhou, Jing Wu, Yipeng Liu, Shuaicheng Liu, and Ce Zhu. "Dast: Data-free substitute training for adversarial attacks." IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), (Oral presentation).

Connect

• Project information
• Google Scholar
• LinkedIn
• Email
• Personal website

Diagram: Improving the Reliability of Mobile AI Framework