In this paper, we examine an important problem of learning neural networks that certifiably meet certain specifications on input-output behaviors. Our strategy is to find an inner approximation of the set of admissible policy parameters, which is convex in a transformed space. To this end, we address the key technical challenge of convexifying the verification condition for neural networks, which is derived by abstracting the nonlinear specifications and activation functions with quadratic constraints. In particular, we propose a reparametrization scheme of the original neural network based on loop transformation, which leads to a convex condition that can be enforced during learning. This theoretical construction is validated in an experiment that specifies reachable sets for different regions of inputs.
Resilient and Effort-Optimal Controller Synthesis under Temporal Logic Specifications
University Mohammed VI Polytechnic（穆罕默德六世 polytechnic 大学）；The Robert Bosch Centre for Cyber-Physical Systems, IISc（印度科学学院罗伯特博世网络物理系统中心）；Newcastle University（纽卡斯尔大学）；Max Planck Institute for Software Systems（马克斯·普朗克软件系统研究所）；University of Birmingham（伯明翰大学）
Cooperative Planning for Coupled Multi-Agent Systems under Timed Temporal Specifications
ACCESS Linnaeus Center, School of Electrical Engineering, KTH Royal Institute of Technology（林奈中心，电气工程学院，皇家理工学院）；KTH Center for Autonomous Systems（自主系统中心）
Politecnico di Milano（米兰理工大学）；Nanyang Technological University（南洋理工大学）；Huazhong University of Science and Technology（华中科技大学）；University of Groningen（格罗宁根大学）
