Publication detail

Optimizing Convolutional Neural Networks for Embedded Systems By Means of Neuroevolution

BADÁŇ, F. SEKANINA, L.

Original Title

Optimizing Convolutional Neural Networks for Embedded Systems By Means of Neuroevolution

Type

conference paper

Language

English

Original Abstract

Automated design methods for convolutional neural networks (CNNs) have recently been developed in order to increase the design productivity. We propose a neuroevolution method capable of evolving and optimizing CNNs with respect to the classification error and CNN complexity (expressed as the number of tunable CNN parameters), in which the inference phase can partly be executed using fixed point operations to further reduce power consumption. Experimental results are obtained with TinyDNN framework and presented using two common image classification benchmark problems - MNIST and CIFAR-10.

Keywords

Evolutionary Algorithm, Convolutional neural network, Neuroevolution,  Embedded Systems, Energy Efficiency

Authors

BADÁŇ, F.; SEKANINA, L.

Released

22. 11. 2019

Publisher

Springer International Publishing

Location

Cham

ISBN

978-3-030-34499-3

Book

Theory and Practice of Natural Computing

Edition

LNCS 11934

Pages from

109

Pages to

121

Pages count

13

URL

https://www.fit.vut.cz/research/publication/12045/

BibTex

@inproceedings{BUT161459,
  author="Filip {Badáň} and Lukáš {Sekanina}",
  title="Optimizing Convolutional Neural Networks for Embedded Systems By Means of Neuroevolution",
  booktitle="Theory and Practice of Natural Computing",
  year="2019",
  series="LNCS 11934",
  pages="109--121",
  publisher="Springer International Publishing",
  address="Cham",
  doi="10.1007/978-3-030-34500-6\{_}7",
  isbn="978-3-030-34499-3",
  url="https://www.fit.vut.cz/research/publication/12045/"
}