User: Guest  Login
If you experience problems opening the document, please try this link.
Original title:
Non-intrusive Model Order Reduction based on Joint Space Sampling and Active Learning
Translated title:
Non-intrusive Model Order Reduction basierend auf Joint Space Sampling und Aktiven Lernens
Author:
Zhuang, Qinyu
Year:
2022
Document type:
Dissertation
Faculty/School:
TUM School of Computation, Information and Technology
Advisor:
Bungartz, Hans-Joachim (Prof. Dr.)
Referee:
Bungartz, Hans-Joachim (Prof. Dr.); Peherstorfer, Benjamin (Prof. Dr.)
Language:
en
Subject group:
DAT Datenverarbeitung, Informatik
TUM classification:
MAT 650; DAT 780
Abstract:
Non-intrusive Model Order Reduction (MOR) is highly demanded by industry. Machine-Learning-based MOR (ML-based MOR) is an important member of non-intrusive MOR. To improve the quality of ML-based Reduced Order Models, we propose a novel method based on Joint Space Sampling and Active Learning. The method can provide high quality training data and reduce the amount of training data required by ML-based MOR.
Translated abstract:
Non-intrusive Model Order Reduction (MOR) wird von der Industrie stark nachgefragt. Machine-Learning-based MOR (ML-based MOR) ist ein wichtiger Bestandteil von non-intrusive MOR. Um die Qualität von ML-based Reduced Order Models zu verbessern, schlagen wir eine neuartige Methode vor, die auf Joint Space Sampling und Active Learning basiert. Das Verfahren kann qualitativ hochwertige Trainingsdaten bereitstellen und die Menge an Trainingsdaten reduzieren, die von ML-based MOR benötigt werden.
WWW:
https://mediatum.ub.tum.de/?id=1686070
Date of submission:
29.08.2022
Oral examination:
19.12.2022
File size:
7077692 bytes
Pages:
171
Urn (citeable URL):
https://nbn-resolving.de/urn/resolver.pl?urn:nbn:de:bvb:91-diss-20221219-1686070-1-1
Last change:
30.12.2022
 BibTeX