معرفی
Weijun Li is a PartTime Lecturer and PhD candidate at Macquarie University, affiliated with the School of Computing within the Faculty of Science and Engineering. His casual academic role indicates a sessional teaching position concurrent with doctoral studies, supported by primary institutional email weijun.li@mq.edu.au and HDR student email weijun.li1@hdr.mq.edu.au.
His research centers on deep learning security vulnerabilities, specializing in adversarial machine learning countermeasures. Key focus areas include backdoor attack mitigation through module switching, post-training model purification, and data leakage analysis in transformer architectures. His work bridges natural language processing security and privacy-preserving techniques, with direct applications in federated learning environments where gradient-based information leakage poses critical risks.
Publications from 2024-2025 demonstrate consistent contributions to AI security: pioneering module substitution for neural network purification, developing defenses against backdoor injections, and exposing gradient leakage mechanisms in transformers. These works establish him as an emerging researcher addressing foundational challenges in trustworthy machine learning systems through rigorous empirical analysis.
