Skip to content

Detail of publication

Citation

M. Bulín and L. Šmídl and J. Švec : Towards Network Simplification for Low-Cost Devices by Removing Synapses . International Conference on Speech and Computer, p. 58-67, Springer, Cham, 2018.

Download PDF

PDF

Additional information


Springer

Abstract

The deployment of robust neural network based models on low-cost devices touches the problem with hardware constraints like limited memory footprint and computing power. This work presents a general method for a rapid reduction of parameters (80–90%) in a trained (DNN or LSTM) network by removing its redundant synapses, while the classification accuracy is not significantly hurt. The massive reduction of parameters leads to a notable decrease of the model’s size and the actual prediction time of on-board classifiers. We show the pruning results on a simple speech recognition task, however, the method is applicable to any classification data.

Detail of publication

Title: Towards Network Simplification for Low-Cost Devices by Removing Synapses
Author: M. Bulín ; L. Šmídl ; J. Švec
Language: English
Date of publication: 18 Sep 2018
Year: 2018
Type of publication: Papers in proceedings of reviewed conferences
Title of journal or book: International Conference on Speech and Computer
Page: 58 - 67
DOI: https://doi.org/10.1007/978-3-319-99579-3_7
ISBN: 978-3-319-99579-3
Publisher: Springer, Cham
Date: 18 Sep 2018 - 22 Sep 2018
/ 2019-02-07 14:45:35 /

Keywords

Pruning synapses, Network simplification, Minimal network structure, Low-cost devices, Speech recognition

BibTeX

@INPROCEEDINGS{MBulin_2018_TowardsNetwork,
 author = {M. Bul\'{i}n and L. \v{S}m\'{i}dl and J. \v{S}vec},
 title = {Towards Network Simplification for Low-Cost Devices by Removing Synapses},
 year = {2018},
 publisher = {Springer, Cham},
 journal = {International Conference on Speech and Computer},
 pages = {58-67},
 ISBN = {978-3-319-99579-3},
 doi = {https://doi.org/10.1007/978-3-319-99579-3_7},
 url = {http://www.kky.zcu.cz/en/publications/MBulin_2018_TowardsNetwork},
}