Wei Du

Papers

Total Citations

H-Index

About

Wei Du is an emerging researcher specializing in adversarial machine learning and 3D computer vision, with a particular focus on the security and robustness of deep learning models applied to three-dimensional data. His most notable work, "3DHacker: Spectrum-based Decision Boundary Generation for Hard-label 3D Point Cloud Attack" (2023), represents a significant contribution to understanding the vulnerabilities of 3D point cloud models — a critical concern as depth sensor technology becomes increasingly prevalent in high-stakes applications such as autonomous driving and robot navigation. In this work, Du and his collaborators tackle the challenging hard-label black-box attack setting, a more realistic and practically relevant threat model compared to white-box approaches, demonstrating that 3D perception systems can be compromised even with severely limited access to model internals. While his publication record is still developing, with his key paper accumulating early citations, his research addresses timely and consequential questions at the intersection of AI security and embodied intelligence. As autonomous systems grow more widespread, Du's investigations into adversarial vulnerabilities position him as a researcher whose work carries meaningful implications for the safety and reliability of next-generation AI-powered technologies.