KIT | KIT-Bibliothek | Impressum | Datenschutz

Smart Data Representations: Impact on the Accuracy of Deep Neural Networks

Neumann, Oliver; Turowski, Marian ORCID iD icon; Ludwig, Nicole; Heidrich, Benedikt; Hagenmeyer, Veit ORCID iD icon; Mikut, Ralf ORCID iD icon

Abstract (englisch):

Deep Neural Networks (DNNs) are able to solve many complex tasks with less engineering effort and better performance. However, these networks often use data for training and evaluation without investigating its representation, i.e. the form of the used data. In the present paper, we analyze the impact of data representations on the performance of DNNs using energy time series forecasting. Based on an overview of exemplary data representations, we select four exemplary data representations and evaluate them using two different DNN architectures and three forecasting horizons on real-world energy time series. The results show that, depending on the forecast horizon, the same data representations can have a positive or negative impact on the accuracy of DNNs.


Verlagsausgabe §
DOI: 10.5445/KSP/1000138532
Cover der Publikation
Zugehörige Institution(en) am KIT Institut für Automation und angewandte Informatik (IAI)
Publikationstyp Proceedingsbeitrag
Publikationsjahr 2021
Sprache Englisch
Identifikator ISBN: 978-3-7315-1131-1
ISSN: 1000138532
KITopen-ID: 1000140589
HGF-Programm 37.12.02 (POF IV, LK 01) Design,Operation & Digitalization of the Future Energy Grids
Erschienen in Proceedings - 31. Workshop Computational Intelligence : Berlin, 25. - 26. November 2021. Hrsg.: H. Schulte; F. Hoffmann; R. Mikut
Veranstaltung 31. Workshop Computational Intelligence (2021), Berlin, Deutschland, 25.11.2021 – 26.11.2021
Verlag KIT Scientific Publishing
Seiten 113-130
Relationen in KITopen
KIT – Die Forschungsuniversität in der Helmholtz-Gemeinschaft
KITopen Landing Page