Abstract: We propose a general attack framework based on evolutionary algorithms to quickly and efficiently generate low-perturbation adversarial samples for 3D point cloud data. Specifically, we ...
Abstract: We expose in a tutorial fashion the mechanisms which underlie the synthesis of optimization algorithms based on dynamic integral quadratic constraints. We reveal how these tools from robust ...