Publication Details

Comparison of FPNNs Models Approximation Capabilities and FPGA Resources Utilization

KRČMA Martin, KOTÁSEK Zdeněk and LOJDA Jakub. Comparison of FPNNs Models Approximation Capabilities and FPGA Resources Utilization. In: Proceedings of IEEE 13th International Conference on Intelligent Computer Communication and Processing. Cluj-Nappoca: IEEE Computer Society, 2017, pp. 125-132. ISBN 978-1-5386-3368-7.
Czech title
Porovnání aproximačních schopností různých typů FPNN a spotřeby zdrojů na FPGA
Type
conference paper
Language
english
Authors
Keywords

ANN, FPNN, FPGA

Abstract

This paper presents the concepts of FPNA and FPNN, used for the approximation of artificial neural networks in FPGAs and introduces derived types of these concepts used by the authors. The process of transformation of a trained artificial neural network to an FPNN is  described. The diagram of the FPGA implementation is presented. The results of experiments determining the approximation capabilities of FPNNs are presented and the FPGA resources utilization are compared.

Published
2017
Pages
125-132
Proceedings
Proceedings of IEEE 13th International Conference on Intelligent Computer Communication and Processing
Conference
IEEE 13th International Conference on Intelligent Computer Communication and Processing, Cluj-Nappoca, RO
ISBN
978-1-5386-3368-7
Publisher
IEEE Computer Society
Place
Cluj-Nappoca, RO
DOI
UT WoS
000417426600015
EID Scopus
BibTeX
@INPROCEEDINGS{FITPUB11507,
   author = "Martin Kr\v{c}ma and Zden\v{e}k Kot\'{a}sek and Jakub Lojda",
   title = "Comparison of FPNNs Models Approximation Capabilities and FPGA Resources Utilization",
   pages = "125--132",
   booktitle = "Proceedings of  IEEE 13th International Conference on Intelligent Computer Communication and Processing",
   year = 2017,
   location = "Cluj-Nappoca, RO",
   publisher = "IEEE Computer Society",
   ISBN = "978-1-5386-3368-7",
   doi = "10.1109/ICCP.2017.8116993",
   language = "english",
   url = "https://www.fit.vut.cz/research/publication/11507"
}
Back to top