Quantifying the Objective Cost of Uncertainty in Complex Dynamical Systems

Real-world problems often involve complex systems that cannot be perfectly modeled or identified, and many engineering applications aim to design operators that can perform reliably in the presence of such uncertainty. In this paper, we propose a novel Bayesian framework for objective-based uncertainty quantification (UQ), which quantifies the uncertainty in a given system based on the expected increase of the operational cost that it induces. This measure of uncertainty, called MOCU (mean objective cost of uncertainty), provides a practical way of quantifying the effect of various types of system uncertainties on the operation of interest. Furthermore, the proposed UQ framework provides a general mathematical basis for designing robust operators, and it can be applied to diverse applications, including robust filtering, classification, and control. We demonstrate the utility and effectiveness of the proposed framework by applying it to the problem of robust structural intervention of gene regulatory networks, an important application in translational genomics.