Sampling in Bayesian Inversion Accelerated by Surrogate Models

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F68145535%3A_____%2F25%3A00642868" target="_blank" >RIV/68145535:_____/25:00642868 - isvavai.cz</a>

Výsledek na webu

<a href="https://doi.org/10.1007/978-3-031-87213-6_44" target="_blank" >https://doi.org/10.1007/978-3-031-87213-6_44</a>

DOI - Digital Object Identifier

<a href="http://dx.doi.org/10.1007/978-3-031-87213-6_44" target="_blank" >10.1007/978-3-031-87213-6_44</a>

Jazyk výsledku

angličtina

Název v původním jazyce

Sampling in Bayesian Inversion Accelerated by Surrogate Models

Popis výsledku v původním jazyce

This chapter explores the acceleration of Bayesian inversion for geotechnical problems using surrogate models, with a focus on neural networks. The study introduces a sampling framework based on the delayed-acceptance Metropolis-Hastings (DAMH) algorithm, which is enhanced with surrogate models to significantly improve efficiency. The framework is applied to a real-world geotechnical inverse problem derived from a tunnel sealing experiment, demonstrating its practical relevance. The chapter compares various surrogate models, including neural networks, radial basis functions, and polynomial chaos approximation, highlighting the superior performance of neural network-based approaches. The results show that refining surrogate models during the sampling process can greatly reduce the number of rejected samples, leading to increased sampling efficiency. The posterior distribution is analyzed and visualized, providing insights into the model's performance and the quality of the surrogate models. The conclusions emphasize the promise of neural network surrogate models for accelerating Bayesian inversion in geotechnical engineering, while noting the importance of careful network configuration and learning method settings.

Název v anglickém jazyce

Sampling in Bayesian Inversion Accelerated by Surrogate Models

Popis výsledku anglicky

This chapter explores the acceleration of Bayesian inversion for geotechnical problems using surrogate models, with a focus on neural networks. The study introduces a sampling framework based on the delayed-acceptance Metropolis-Hastings (DAMH) algorithm, which is enhanced with surrogate models to significantly improve efficiency. The framework is applied to a real-world geotechnical inverse problem derived from a tunnel sealing experiment, demonstrating its practical relevance. The chapter compares various surrogate models, including neural networks, radial basis functions, and polynomial chaos approximation, highlighting the superior performance of neural network-based approaches. The results show that refining surrogate models during the sampling process can greatly reduce the number of rejected samples, leading to increased sampling efficiency. The posterior distribution is analyzed and visualized, providing insights into the model's performance and the quality of the surrogate models. The conclusions emphasize the promise of neural network surrogate models for accelerating Bayesian inversion in geotechnical engineering, while noting the importance of careful network configuration and learning method settings.

Druh

D - Stať ve sborníku

CEP obor

—

OECD FORD obor

10102 - Applied mathematics

Projekt

—

Návaznosti

I - Institucionalni podpora na dlouhodoby koncepcni rozvoj vyzkumne organizace

Rok uplatnění

2025

Kód důvěrnosti údajů

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

Název statě ve sborníku

Inverse Problems: Modelling and Simulation : Extended Abstracts of the IPMS Conference 2024

ISBN

978-3-031-87212-9

ISSN

—

e-ISSN

—

Počet stran výsledku

7

Strana od-do

365-371

Název nakladatele

Springer

Místo vydání

Cham

Místo konání akce

Paradise Bay Resort Hotel, Malta

Datum konání akce

26. 5. 2024

Typ akce podle státní příslušnosti

WRD - Celosvětová akce

Kód UT WoS článku

—