Yinpeng Dong

Papers

Total Citations

H-Index

About

Yinpeng Dong is a leading researcher in adversarial machine learning and AI security, with a focus on the robustness of deep learning systems in real-world applications. His work critically examines vulnerabilities in 3D perception systems—essential for autonomous driving and robotics—demonstrating how state-of-the-art point cloud classifiers can be fooled by carefully crafted adversarial attacks. In his highly cited 2023 paper, "The Art of Defense: Letting Networks Fool the Attacker," Dong introduces novel defensive strategies that turn the tables on adversaries, showing how networks can be trained to mislead attackers rather than simply resist them. This work, which has already garnered 19 citations, represents a paradigm shift in thinking about AI security. Beyond this, Dong's broader contributions include pioneering methods for generating robust adversarial examples and developing defenses that maintain performance under attack. His research has significant implications for deploying trustworthy AI in safety-critical domains, making him a key voice in the ongoing effort to build more resilient machine learning systems.