On a Unified Framework for Deterministic & Stochastic Treatment of Identification Problems
Alexander B. Kurzhanski, Masahiro Tanaka · 1989
This paper deals with the conventional problem of identifying a matrix parameter on the basis of observations corrupted by an uncertainty in the measurements. Recalling two basic approaches to this problem -- the stochastic scheme when the error in observation is treated as a Gaussian noise and the deterministic approach with only a set-membership description of the unknown variables, the paper indicates the connections and interactions in the techniques involved in the respective solutions.