TL;DR人体运动预测的敌对攻击实验表明,类似于图像分类中的初始阶段,针对 GCN 和基于 MLP 的多回归模型进行的广泛实验表明模型在较低程度的扰动下容易受到攻击,以及简单的 3D 变换能够影响模型的性能。
Abstract
human motion prediction is still an open problem, which is extremely
important for autonomous driving and safety applications. Although there are
great advances in this area, the widely studied topic of adversarial atta