Local Riemannian geometry of model manifolds and its implications for practical parameter identifiability

Abstract:

When non-linear models are fitted to experimental data, parameter estimates can be poorly constrained albeit being identifiable in principle. This means that along certain paths in parameter space, the log-likelihood does not exceed a given statistical threshold but remains bounded. This situation, denoted as practical non-identifiability, can be detected by Monte Carlo sampling or by systematic scanning using the profile likelihood method. In contrast, any method based on a Taylor expansion of the log-likelihood around the optimum, e.g., parameter uncertainty estimation by the Fisher Information Matrix, reveals no information about the boundedness at all. In this work, we present a geometric approach, approximating the original log-likelihood by geodesic coordinates of the model manifold. The Christoffel Symbols in the geodesic equation are fixed to those obtained from second order model sensitivities at the optimum. Based on three exemplary non-linear models we show that the information about the log-likelihood bounds and flat parameter directions can already be contained in this local information. Whereas the unbounded case represented by the Fisher Information Matrix is embedded in the geometric framework as vanishing Christoffel Symbols, non-vanishing constant Christoffel Symbols prove to define prototype non-linear models featuring boundedness and flat parameter directions of the log-likelihood. Finally, we investigate if those models could allow to approximate and replace computationally expensive objective functions originating from non-linear models by a surrogate objective function in parameter estimation problems.

Citation: PLoS ONE 14(6) : e0217837

Date Published: 3rd Jun 2019

Authors: Daniel Lill, Jens Timmer, Daniel Kaschek

Help
help Creator
Citation
Lill, D., Timmer, J., & Kaschek, D. (2019). Local Riemannian geometry of model manifolds and its implications for practical parameter identifiability. PLOS ONE, 14(6), e0217837. http://doi.org/10.1371/journal.pone.0217837
Activity

Views: 108

Created: 24th Jul 2019 at 12:21

Last updated: 24th Jul 2019 at 12:23

help Attributions

None

Related items

Powered by
(v.1.9.0)
Copyright © 2008 - 2019 The University of Manchester and HITS gGmbH