Active Learning for Efficient Rare Event Probability Estimation and Sensitivity Analyses in Highly Nonlinear Systems

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F00216305%3A26110%2F24%3APU155250" target="_blank" >RIV/00216305:26110/24:PU155250 - isvavai.cz</a>

Výsledek na webu

<a href="http://dx.doi.org/10.1007/978-3-031-60271-9_30" target="_blank" >http://dx.doi.org/10.1007/978-3-031-60271-9_30</a>

DOI - Digital Object Identifier

<a href="http://dx.doi.org/10.1007/978-3-031-60271-9_30" target="_blank" >10.1007/978-3-031-60271-9_30</a>

Jazyk výsledku

angličtina

Název v původním jazyce

Active Learning for Efficient Rare Event Probability Estimation and Sensitivity Analyses in Highly Nonlinear Systems

Popis výsledku v původním jazyce

This paper presents a robust method for rare event probability estimation in highly nonlinear systems. Utilizing a nearest-neighbor approximation of the true performance function and an adaptively extended experimental design, we introduce a simple yet effective active learning function. This function dynamically balances global exploration and local exploitation through sequential adaptive selection of points from the input domain. The resulting surrogate model, refined based on distances, serves the dual purpose of estimating failure probability and selecting optimal candidates for further model evaluations. Our adaptive design supports accurate real-time estimation of failure probability and failure probability sensitivity to individual variables, especially in cases of non-smooth or highly nonlinear functions. Even in scenarios with smooth functions, our method outperforms existing approaches utilizing the function gradients in estimation accuracy for a given computational budget. The adaptively constructed surrogate model excels in handling intricate failure surfaces, multiple design points, and systems with bifurcations. This approach is particularly suitable for random vectors with small to moderate dimensions.

Název v anglickém jazyce

Active Learning for Efficient Rare Event Probability Estimation and Sensitivity Analyses in Highly Nonlinear Systems

Popis výsledku anglicky

This paper presents a robust method for rare event probability estimation in highly nonlinear systems. Utilizing a nearest-neighbor approximation of the true performance function and an adaptively extended experimental design, we introduce a simple yet effective active learning function. This function dynamically balances global exploration and local exploitation through sequential adaptive selection of points from the input domain. The resulting surrogate model, refined based on distances, serves the dual purpose of estimating failure probability and selecting optimal candidates for further model evaluations. Our adaptive design supports accurate real-time estimation of failure probability and failure probability sensitivity to individual variables, especially in cases of non-smooth or highly nonlinear functions. Even in scenarios with smooth functions, our method outperforms existing approaches utilizing the function gradients in estimation accuracy for a given computational budget. The adaptively constructed surrogate model excels in handling intricate failure surfaces, multiple design points, and systems with bifurcations. This approach is particularly suitable for random vectors with small to moderate dimensions.

Druh

D - Stať ve sborníku

CEP obor

—

OECD FORD obor

20102 - Construction engineering, Municipal and structural engineering

Projekt

<a href="/cs/project/GA24-10892S" target="_blank" >GA24-10892S: Strojové učení pro víceúrovňové modelování prostorové variability a trhlin pro zajištění udržitelnosti betonových konstrukcí</a><br>

Návaznosti

P - Projekt vyzkumu a vyvoje financovany z verejnych zdroju (s odkazem do CEP)

Rok uplatnění

2024

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Název statě ve sborníku

Lecture Notes in Civil Engineering

ISBN

978-3-031-60271-9

ISSN

2366-2557

e-ISSN

—

Počet stran výsledku

10

Strana od-do

324-333

Název nakladatele

SPRINGER INTERNATIONAL PUBLISHING AG

Místo vydání

CHAM

Místo konání akce

Guimarães

Datum konání akce

8. 5. 2024

Typ akce podle státní příslušnosti

WRD - Celosvětová akce

Kód UT WoS článku

001323733800032