2006
DOI: 10.1088/0026-1394/43/5/009
|View full text |Cite
|
Sign up to set email alerts
|

The expression of uncertainty in non-linear parameter estimation

Abstract: We made an analysis of uncertainty propagation in parameter estimation when the cost function minimization leads to a set of non-linear equations. In our study, which concerns an uncertainty-propagation law independent of any specific minimization algorithm, we review the relationship between the Hessian of the cost function and the estimate uncertainty, as well as the reasons supporting their identification. We show that the uncertainty expressed on this basis could lead to accuracy overestimation.

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
2
2
1

Citation Types

0
21
0

Year Published

2007
2007
2024
2024

Publication Types

Select...
7
1

Relationship

0
8

Authors

Journals

citations
Cited by 21 publications
(21 citation statements)
references
References 4 publications
0
21
0
Order By: Relevance
“…We used MATLAB to analyze data by nonlinear least squares parameter fitting (Supplementary Material, see Additional File 2 for codebase). Uncertainties of parameters were measured according to the method of Balsamo et al 17 To disaggregate mean study group values to individual subjects, we used Monte Carlo sampling of the immune state. The immune trajectories were calculated individually for each sampled subject and finally aggregated back to their respective synthetic study groups to compare predicted to observed outcomes.…”
Section: Methodsmentioning
confidence: 99%
“…We used MATLAB to analyze data by nonlinear least squares parameter fitting (Supplementary Material, see Additional File 2 for codebase). Uncertainties of parameters were measured according to the method of Balsamo et al 17 To disaggregate mean study group values to individual subjects, we used Monte Carlo sampling of the immune state. The immune trajectories were calculated individually for each sampled subject and finally aggregated back to their respective synthetic study groups to compare predicted to observed outcomes.…”
Section: Methodsmentioning
confidence: 99%
“…Since the explicit solution is only available for very special cases, the parameters of the calibration function are usually determined by iterative numerical procedures that minimise the criterion function (8) based on the total sum of the weighted squares of the residues.…”
Section: Linear Comparative Calibration Modelmentioning
confidence: 99%
“…[13], [21], [26], [23], [22], [24], and [9]. The estimated covariance matrix is usually based on the information matrices or their observed version based on the calculated Hessian matrix, although simplifications are possible, see for example [8] and [17] for a detailed discussion of the available approaches and comparisons in nonlinear measurement models.…”
Section: Linear Comparative Calibration Modelmentioning
confidence: 99%
“…Validation of measuring methods is one of fastest growing fields of metrology with important applications in production engineering and, in a broader sense, the production of various machines and devices. According to [2,[4][5][6][7][8][9], there is a pressing need to develop, especially in the field of coordinate metrology, new measuring methods with the mandatory assessment of their accuracy and, above all, an effective model for their validation. Such model would call for the development of an algorithm used for validation of measuring and calibration methods, both at calibration and research laboratories, as well as industrial laboratories supervising production processes.…”
Section: Introductionmentioning
confidence: 99%