Jinyuan Jia

Papers

Total Citations

H-Index

About

Jinyuan Jia is a researcher specializing in adversarial machine learning and trustworthy AI, with a particular focus on the security and robustness of three-dimensional perception systems. His most recognized contribution, **PointGuard: Provably Robust 3D Point Cloud Classification** (2021), addresses critical vulnerabilities in 3D point cloud classifiers — systems that underpin safety-critical technologies such as autonomous vehicles and robotic manipulation. By demonstrating that adversarial attackers can manipulate classifier predictions through carefully crafted perturbations, Jia and his collaborators identified a significant threat surface in modern AI pipelines. More importantly, PointGuard offers *provable* robustness guarantees, moving beyond empirical defenses to mathematically certifiable protections — a meaningful advance in the field. The work has accumulated over 60 citations, reflecting strong uptake within the adversarial robustness and 3D vision communities. Jia's research sits at the intersection of machine learning security, geometric deep learning, and real-world AI deployment, making his work highly relevant to researchers and practitioners building reliable perception systems. Students interested in certified defenses, point cloud processing, or AI safety will find his contributions a valuable entry point into these rapidly evolving areas.