Publication Details

Detecting hard synapses faults in artificial neural networks

KRČMA Martin, KOTÁSEK Zdeněk and LOJDA Jakub. Detecting hard synapses faults in artificial neural networks. In: 20th IEEE Latin American Test Symposium (LATS 2019). Santiago de Chile: IEEE Computer Society, 2019, pp. 1-6. ISBN 978-1-7281-1756-0.
Czech title
Detekce trvalých poruch v synapsích neuronových sítí
Type
conference paper
Language
english
Authors
Krčma Martin, Ing. (DCSY FIT BUT)
Kotásek Zdeněk, doc. Ing., CSc. (DCSY FIT BUT)
Lojda Jakub, Ing., Ph.D. (DCSY FIT BUT)
Keywords

artificial neural networks, hard faults, faults detection, fault tolerance

Abstract

This paper presents the concepts of detecting hard faults in artificial neural network synapses using the modification of the neural network settings. The core of this work is based on weights values modification and inserting the chosen testing data when comparing the neural network output to the known valid results. The paper also discuss the problem of neural network output saturation and provide experiments on influence of the neural network settings to the problem in this regard.

Published
2019
Pages
1-6
Proceedings
20th IEEE Latin American Test Symposium (LATS 2019)
Conference
IEEE Latin American Test Symposium, Hotel Fundador, Santiago de Chile, CL
ISBN
978-1-7281-1756-0
Publisher
IEEE Computer Society
Place
Santiago de Chile, CL
DOI
10.1109/LATW.2019.8704637
UT WoS
000469850000035
EID Scopus
2-s2.0-85065958948
BibTeX 
@INPROCEEDINGS{FITPUB11876,
   author = "Martin Kr\v{c}ma and Zden\v{e}k Kot\'{a}sek and Jakub Lojda",
   title = "Detecting hard synapses faults in artificial neural networks",
   pages = "1--6",
   booktitle = "20th IEEE Latin American Test Symposium (LATS 2019)",
   year = 2019,
   location = "Santiago de Chile, CL",
   publisher = "IEEE Computer Society",
   ISBN = "978-1-7281-1756-0",
   doi = "10.1109/LATW.2019.8704637",
   language = "english",
   url = "https://www.fit.vut.cz/research/publication/11876"
}
Back to top