CRC 1294/1: Nonlinear Statistical Inverse Problems With Random Observations (SP A04)
Abstract
The project deals with nonlinear statistical inverse problems; the goal is to estimate from random observations the functional relation between observable covariates and intrinsic (unobservable) parameters of a system whose output is observed. The system itself is known (up to the intrinsic parameters) through a specific model, e.g. a differential equation. The main objective is to design suitable non-parametric estimators using a reproducing kernel ansatz, i.e., nonlinear regularized maximum likelihood estimators, and to analyze their theoretical performance. Mechanistic modelling approaches from pharmacokinetics serve as a specific application. Questions relating to efficient computation of the estimate as well as associated quantification of uncertainty will be addressed.
Principal investigators
Participating external organisations
Financer
DFG Collaborative Research Centre
Duration of project
Start date: 07/2017
End date: 06/2021