Computer Science > Machine Learning
[Submitted on 14 May 2021 (this version), latest version 26 Oct 2021 (v2)]
Title:Partitioned Active Learning for Heterogeneous Systems
View PDFAbstract:Cost-effective and high-precision surrogate modeling is a cornerstone of automated industrial and engineering systems. Active learning coupled with Gaussian process (GP) surrogate modeling is an indispensable tool for demanding and complex systems, while the existence of heterogeneity in underlying systems may adversely affect the modeling process. In order to improve the learning efficiency under the regime, we propose the partitioned active learning strategy established upon partitioned GP (PGP) modeling. Our strategy seeks the most informative design point for PGP modeling systematically in twosteps. The global searching scheme accelerates the exploration aspect of active learning by investigating the most uncertain design space, and the local searching exploits the active learning criterion induced by the local GP model. We also provide numerical remedies to alleviate the computational cost of active learning, thereby allowing the proposed method to incorporate a large amount of candidates. The proposed method is applied to numerical simulation and real world cases endowed with heterogeneities in which surrogate models are constructed to embed in (i) the cost-efficient automatic fuselage shape control system; and (ii) the optimal design system of tribocorrosion-resistant alloys. The results show that our approach outperforms benchmark methods.
Submission history
From: Cheolhei Lee [view email][v1] Fri, 14 May 2021 02:05:31 UTC (2,519 KB)
[v2] Tue, 26 Oct 2021 15:22:41 UTC (4,860 KB)
References & Citations
Bibliographic and Citation Tools
Bibliographic Explorer (What is the Explorer?)
Litmaps (What is Litmaps?)
scite Smart Citations (What are Smart Citations?)
Code, Data and Media Associated with this Article
alphaXiv (What is alphaXiv?)
CatalyzeX Code Finder for Papers (What is CatalyzeX?)
DagsHub (What is DagsHub?)
Gotit.pub (What is GotitPub?)
Hugging Face (What is Huggingface?)
Papers with Code (What is Papers with Code?)
ScienceCast (What is ScienceCast?)
Demos
Recommenders and Search Tools
Influence Flower (What are Influence Flowers?)
Connected Papers (What is Connected Papers?)
CORE Recommender (What is CORE?)
IArxiv Recommender
(What is IArxiv?)
arXivLabs: experimental projects with community collaborators
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.