Imperial College London
Portrait Picture

ProfessorRuthMisener

Faculty of Engineering, Department of Computing

Professor in Computational Optimisation
 
 
 
//

Contact

 

+44 (0)20 7594 8315r.misener Website CV

 
 
//

Location

 

379Huxley BuildingSouth Kensington Campus

//

Summary

 

Publications

Citation

BibTex format

@unpublished{Olofsson:2021,
author = {Olofsson, S and Schultz, ES and Mhamdi, A and Mitsos, A and Deisenroth, MP and Misener, R},
publisher = {arXiv},
title = {Design of dynamic experiments for black-box model discrimination},
url = {http://arxiv.org/abs/2102.03782v1},
year = {2021}
}
Download

RIS format (EndNote, RefMan)

TY  - UNPB
AB  - Diverse domains of science and engineering require and use mechanisticmathematical models, e.g. systems of differential algebraic equations. Suchmodels often contain uncertain parameters to be estimated from data. Consider adynamic model discrimination setting where we wish to chose: (i) what is thebest mechanistic, time-varying model and (ii) what are the best model parameterestimates. These tasks are often termed modeldiscrimination/selection/validation/verification. Typically, several rivalmechanistic models can explain data, so we incorporate available data and alsorun new experiments to gather more data. Design of dynamic experiments formodel discrimination helps optimally collect data. For rival mechanistic modelswhere we have access to gradient information, we extend existing methods toincorporate a wider range of problem uncertainty and show that our proposedapproach is equivalent to historical approaches when limiting the types ofconsidered uncertainty. We also consider rival mechanistic models as dynamicblack boxes that we can evaluate, e.g. by running legacy code, but wheregradient or other advanced information is unavailable. We replace theseblack-box models with Gaussian process surrogate models and thereby extend themodel discrimination setting to additionally incorporate rival black-box model.We also explore the consequences of using Gaussian process surrogates toapproximate gradient-based methods.
AU  - Olofsson,S
AU  - Schultz,ES
AU  - Mhamdi,A
AU  - Mitsos,A
AU  - Deisenroth,MP
AU  - Misener,R
PB  - arXiv
PY  - 2021///
TI  - Design of dynamic experiments for black-box model discrimination
UR  - http://arxiv.org/abs/2102.03782v1
UR  - http://hdl.handle.net/10044/1/86940
ER  -
Download