Motasem Alfarra

King Abdullah University of Science and Technology

Papers

Total Citations

H-Index

About

Motasem Alfarra is a researcher at the forefront of trustworthy 3D computer vision, with a primary focus on certifying the robustness of deep learning models against real-world spatial deformations. His major contribution, the work “3DeformRS: Certifying Spatial Deformations on Point Clouds” (2022), addresses a critical vulnerability in 3D perception systems used in autonomous driving and surgical robotics. By proposing a method to provide certified robustness against non-rigid transformations—such as twisting, bending, and stretching of point clouds—Alfarra bridges the gap between theoretical safety guarantees and practical deployment. This pioneering approach has already garnered 5 citations, signaling its importance in the emerging field of certified adversarial robustness for 3D data. His research is particularly notable for tackling a previously underexplored threat model, moving beyond traditional pixel-level attacks to consider physically plausible deformations. Alfarra’s work is essential reading for students and researchers interested in reliable 3D perception, formal verification of neural networks, and building safety-critical autonomous systems that can withstand real-world perturbations.