On optimal designs for non-regular models
arXiv:1802.10024 · doi:10.1214/18-AOS1780
Abstract
Classically, Fisher information is the relevant object in defining optimal experimental designs. However, for models that lack certain regularity, the Fisher information does not exist and, hence, there is no notion of design optimality available in the literature. This article seeks to fill the gap by proposing a so-called Hellinger information, which generalizes Fisher information in the sense that the two measures agree in regular problems, but the former also exists for certain types of non-regular problems. We derive a Hellinger information inequality, showing that Hellinger information defines a lower bound on the local minimax risk of estimators. This provides a connection between features of the underlying model---in particular, the design---and the performance of estimators, motivating the use of this new Hellinger information for non-regular optimal design problems. Hellinger optimal designs are derived for several non-regular regression problems, with numerical results empirically demonstrating the efficiency of these designs compared to alternatives.
arXiv comment23 pages of text; 3 figures; 11 pages of supplementary material
References in corpus (7)
- Computing exact -optimal designs by mixed integer second-order cone programming
- D-optimal designs via a cocktail algorithm
- Optimal discrimination designs
- Support points of locally optimal designs for nonlinear models with two parameters
- A note on the de la Garza phenomenon for locally optimal designs
- Identifying locally optimal designs for nonlinear models: A simple extension with profound consequences
- Complete classes of designs for nonlinear regression models and principal representations of moment spaces