Unifying Qualitative and Quantitative Safety Verification of DNN-Controlled Systems

arXiv:2404.01769v1 Announce Type: new
Abstract: The rapid advance of deep reinforcement learning techniques enables the oversight of safety-critical systems through the utilization of Deep Neural Networks (DNNs). This underscores the pressing need to promptly establish certified safety guarantees for such DNN-controlled systems. Most of the existing verification approaches rely on qualitative approaches, predominantly employing reachability analysis. However, qualitative verification proves inadequate for DNN-controlled systems as their behaviors exhibit stochastic tendencies when operating in open and adversarial environments. In this paper, …

abstract advance arxiv cs.lg dnn networks neural networks oversight quantitative reinforcement reinforcement learning safety safety-critical systems through type verification