Hanxiao Tan

Papers

Total Citations

H-Index

About

Hanxiao Tan is a researcher working at the intersection of explainable artificial intelligence (XAI) and 3D deep learning, with a particular focus on making neural networks more interpretable in safety-critical domains such as autonomous driving and robotics. Tan's most notable contribution centers on developing surrogate model-based explainability methods for point cloud neural networks — a timely and impactful research direction given the rapid adoption of 3D sensors like LiDAR in real-world systems. Point clouds, as the raw output of these sensors, demand specialized neural architectures, yet understanding *why* these networks make particular decisions had remained largely unexplored prior to Tan's work. By addressing this gap, Tan has helped lay foundational groundwork for trustworthy AI in perception systems where reliability is paramount. The research has accumulated 28 citations across its publications, demonstrating growing community recognition of its importance. Tan's work is especially valuable for students and practitioners seeking to deploy explainable 3D perception models in autonomous systems, offering both theoretical grounding and practical methodology for interpreting complex point cloud neural networks.