Document worth reading: “Verification for Machine Learning, Autonomy, and Neural Networks Survey”
This survey presents a top level view of verification methods for autonomous packages, with a focus on safety-critical autonomous cyber-physical packages (CPS) and subcomponents thereof. Autonomy in CPS is enabling by present advances in artificial intelligence (AI) and machine finding out (ML) through approaches comparable to deep neural networks (DNNs), embedded in so-called finding out enabled parts (LECs) that accomplish duties from classification to manage. Recently, the formal methods and formal verification group has developed methods to characterize behaviors in these LECs with eventual targets of formally verifying specs for LECs, and this textual content presents a survey of a lot of these present approaches. Verification for Machine Learning, Autonomy, and Neural Networks Survey