Principles and Metrics for Curating Large Engineering Simulation Datasets for Machine Learning
Journal of Computing and Information Science in Engineering · 01.12.2025 · 25 (12)
DOI: 10.1115/1.4070033 ↗Kurz gefasst
Fachartikel in der Zeitschrift Journal of Computing and Information Science in Engineering, online erschienen am 02.12.2025. Verfasst von Jami J. Shah und 8 weiteren Autor:innen. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 57 Literaturangaben. Gefördert von: Directorate for Engineering.
Themen: Künstliche Intelligenz in der Pflege, Simulation in der Pflegeausbildung
Abstract (Originalsprache)
It is time to talk about data in its own right, not just its usage! Machine learning applications are using a wide variety of data sources, some real, such as data collected by sensors and cameras in driving, and some artificial, such as data generated through numerical simulations. The latter mode has been gaining rapid popularity for engineering design and analysis. We are now seeing the beginnings of publicly shared datasets. Thus, the quality and efficacy of such data need to be considered before their use. In this article, we attempt to outline systematic principles and quantifiable quality and efficacy metrics based on insights gained collectively from both data curation projects and usage of large engineering datasets. Specifically, this article addresses issues related to generating BIG datasets from computer-aided design (CAD) and finite element analysis (FEA): granularity and modality, applicable to both input and output data; and variety and balance, applicable to input data; also, efficacy for machine learning (ML). Generation of Big Data by simulation requires the use of commercial CAD and finite element packages, which poses multiple challenges: automation, integration and balancing sample variants. We propose parametric variety and balance matrices to study over or under representation of input attributes. Even if we have a large dataset with good balance, it may not be suitable for ML if we do not see significant variation in response or performance variables. Several data generation, curation, and utilization case studies are included in a variety of domains (aero, structural, thermal, manufacturing) and usage of metrics is demonstrated.
Förderung
- Directorate for Engineering
Zitieren
APA
Shah, J. J., Ramnath, S., Menzel, S., Rios, T., Kocer, F., Whalen, E., Pajot, J., Adrian, A., & Kumar, P. (2025). Principles and Metrics for Curating Large Engineering Simulation Datasets for Machine Learning. Journal of Computing and Information Science in Engineering, 25(12). https://doi.org/10.1115/1.4070033
Vancouver
Shah JJ, Ramnath S, Menzel S, Rios T, Kocer F, Whalen E, et al. Principles and Metrics for Curating Large Engineering Simulation Datasets for Machine Learning. Journal of Computing and Information Science in Engineering. 2025;25(12). doi:10.1115/1.4070033
BibTeX
@article{Shah2025,
author = {Shah, Jami J. and Ramnath, Satchit and Menzel, Stefan and Rios, Thiago and Kocer, Fatma and Whalen, Eamon and Pajot, Joseph and Adrian, Alex and Kumar, Prakash},
title = {Principles and Metrics for Curating Large Engineering Simulation Datasets for Machine Learning},
journal = {Journal of Computing and Information Science in Engineering},
year = {2025},
volume = {25},
number = {12},
doi = {10.1115/1.4070033}
}
EUVIMED