|
Activity Number:
|
582
- Nonparametric Methods for Statistical Inference
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Wednesday, July 31, 2019 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Section on Nonparametric Statistics
|
|
Abstract #305192
|
Presentation
|
|
Title:
|
Estimation of an Improved Surrogate Model in Uncertainty Quantification by Neural Networks
|
|
Author(s):
|
Sebastian Kersting* and Michael Kohler and Benedict Götz
|
|
Companies:
|
TU Darmstadt and Technische Universitaet Darmstadt and TU Darmstadt
|
|
Keywords:
|
curse of dimensionality;
density estimation;
imperfect models;
neural networks;
surrogate models;
uncertainty quantification
|
|
Abstract:
|
Quantification of uncertainty of a technical system is often based on a surrogate model of a corresponding simulation model. In any application the simulation model will not describe the reality perfectly, and consequently the surrogate model will be imperfect. We propose a method to combine observed data from the technical system with simulated data from the imperfect simulation model in order to estimate an improved surrogate model consisting of multi-layer feedforward neural networks, and show that under suitable assumptions this estimate is able to circumvent the curse of dimensionality. Based on this improved surrogate model we show a rate of convergence result for density estimates. The practical usefulness of the newly proposed estimates is demonstrated by using them to predict the uncertainty of a lateral vibration attenuation system with piezo-elastic supports.
|
Authors who are presenting talks have a * after their name.
