On the Fundamental Tautology of Validating Data-Driven Models and Simulations

John Michopoulos and Sam Lambrakos

Materials Science and Technology Division,U.S. Naval Research Laboratory, Washington, DC. 20375, U.S.A
john.michopoulos@nrl.navy.mil
lambrakos@nrl.navy.mil

Abstract. Recent advances in Dynamic Data Driven Application Systems (DDDAS) facilitated by the present level of computational technologies, as well as advances in data-driven modeling and simulation, impose the need for a critical evaluation of paradigms underlying Qualification, Validation and Verification (QV&V). This paper discusses the fundamental irrelevance of conventional validation procedures with respect to data-driven models and simulations. This inherent property of data-driven models and simulations makes the data-driven approaches extremely desirable from a reliability perspective. An informal comparison of the logical flow of traditional and evolved QV&V demonstrates the tautological nature of data-driven model validation. A brief epistemological review of the origins of traditional and evolved QV&V is also presented.

LNCS 3515, pp. 738-745.

Last modified: