download0 view172
twitter facebook

공공누리This item is licensed Korea Open Government License

Title
Development of a simulation result management and prediction system using machine learning techniques
Author(s)
이기용서영균조금원
Publication Year
2017-12-02
Abstract
Simulations are widely used in various fields of computationalscience and engineering. As IT technology advances, the complexity andaccuracy requirements of the simulations are increasingly rising up,accordingly escalating their execution cost as well. Nevertheless, it appears thatthe community has not yet paid much attention to the reuse of previouslyobtained simulation results to improve the performance of the execution oflater requested simulations. In this regard, we propose a novel simulationservice system that can utilise the results of previously executed simulationsand thus improve the performance of later simulations. The proposed systemcan not only convert completed simulation results into a standard form andthen store them into a NoSQL database for efficient retrieval, but also predictthe result of a requested simulation using machine learning techniques withoutactual simulations. We demonstrate that the proposed system achieved verylow error prediction rates only up to 7.4% from 0.9%.
Keyword
simulation service system; simulation result prediction; machine learning; simulation result reuse
Journal Title
International Journal of Data Mining and Bioinformatics
Citation Volume
19
ISSN
1748-5681
Files in This Item:
There are no files associated with this item.
Appears in Collections:
7. KISTI 연구성과 > 학술지 발표논문
URI
https://repository.kisti.re.kr/handle/10580/14697
http://www.ndsl.kr/ndsl/search/detail/article/articleSearchResultDetail.do?cn=NART90356784
Export
RIS (EndNote)
XLS (Excel)
XML

Browse