Certified Robust Control under Adversarial Perturbations
Certified Robust Control under Adversarial Perturbations
复制标题
DOI：
10.23919/acc55779.2023.10155878
复制
发表时间：
2023-02
期刊：
影响因子：
--
通讯作者：
Jinghan Yang;Hunmin Kim;Wenbin Wan;N. Hovakimyan;Yevgeniy Vorobeychik
中科院分区：
文献类型：
--
作者：
Jinghan Yang;Hunmin Kim;Wenbin Wan;N. Hovakimyan;Yevgeniy Vorobeychik
自主系统越来越依赖机器学习技术，将高维原始输入转化为预测，然后用于决策和控制。然而，恶意操纵这类输入往往很容易，结果就是预测。虽然已经提出了有效的技术来证明预测对对抗性输入扰动的稳健性，但这种技术已经从使用预测的下游控制系统中分离出来。我们提出了第一种方法，将关于原始输入摄动的预测的稳健性证明与鲁棒控制相结合，以获得控制对对抗性输入摄动的经证明的鲁棒性。我们使用一个自适应车辆控制的案例来说明我们的方法，并通过广泛的实验展示了所得到的端到端证书的价值。
Autonomous systems increasingly rely on machine learning techniques to transform high-dimensional raw inputs into predictions that are then used for decision-making and control. However, it is often easy to maliciously manipulate such inputs and, as a result, predictions. While effective techniques have been proposed to certify the robustness of predictions to adversarial input perturbations, such techniques have been disembodied from control systems that make downstream use of the predictions. We propose the first approach for composing robustness certification of predictions with respect to raw input perturbations with robust control to obtain certified robustness of control to adversarial input perturbations. We use a case study of adaptive vehicle control to illustrate our approach and show the value of the resulting end-to-end certificates through extensive experiments.