A Framework for the Analysis of Deep Neural Networks in Aerospace applications using Bayesian Statistics

Deep Neural Networks (DNNs) have gained tremendous popularity in many application areas over the recent years. Safety-critical applications as found in the aerospace domain require that the behavior of the DNN is validated and tested rigorously for system safety.In this paper, we present a framework to support testing of DNNs. Our framework employs techniques from statistical modeling and active learning to effectively generate test cases for DNNs used in Aerospace systems and also supports a comparison between different DNNs. In this paper, we will describe our statistical framework, the algorithms for model construction and the metric guiding the test case generation process.We will present a case study on a physics-based Deep recurrent residual neural network (DR-RNN), which has been trained to emulate the aerodynamics behavior of a Boeing 747-100 aircraft.