Papers
2
Total Citations
3
H-Index
1
About
Qisong He is a rising researcher in the field of adversarial machine learning, with a focused expertise in the security and robustness of 3D point cloud processing. His major contributions center on developing efficient and imperceptible adversarial attacks against 3D deep learning models, a critical area for autonomous systems and computer vision. He is best known for his pioneering work on the "Eidos" framework, introduced in 2024 and expanded in 2025, which demonstrates how to craft adversarial examples that are both computationally lightweight and visually undetectable to human observers. The expanded version, "Eidos Revisited," has already garnered early citations, signaling its growing influence in the community. By tackling the unique challenges of 3D data—such as sparsity and irregular structure—He's research provides foundational insights into the vulnerabilities of point cloud classifiers, directly informing the development of more secure and reliable AI systems for real-world applications like self-driving cars and robotics.
Research Focus
Key Achievements
Top Papers
- 1
- 2Eidos: Efficient, Imperceptible Adversarial 3D Point Clouds1 citations · 2024