Studying Stability of Different Convolutional Neural Networks Against Additive Noise

Hamed H. Aghdam, Elnaz J. Heravi, Domenec Puig

2017

Abstract

Understanding internal process of ConvNets is commonly done using visualization techniques. However, these techniques do not usually provide a tool for estimating stability of a ConvNet against noise. In this paper, we show how to analyze a ConvNet in the frequency domain. Using the frequency domain analysis, we show the reason that a ConvNet might be sensitive to a very low magnitude additive noise. Our experiments on a few ConvNets trained on different datasets reveals that convolution kernels of a trained ConvNet usually pass most of the frequencies and they are not able to effectively eliminate the effect of high frequencies.They also show that a convolution kernel with more concentrated frequency response is more stable against noise. Finally, we illustrate that augmenting a dataset with noisy images can compress the frequency response of convolution kernels.

Paper Citation

in Harvard Style

Aghdam H., J. Heravi E. and Puig D. (2017). Studying Stability of Different Convolutional Neural Networks Against Additive Noise . In Proceedings of the 12th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications - Volume 5: VISAPP, (VISIGRAPP 2017) ISBN 978-989-758-226-4, pages 362-369. DOI: 10.5220/0006200003620369

in Bibtex Style

@conference{visapp17,
author={Hamed H. Aghdam and Elnaz J. Heravi and Domenec Puig},
title={Studying Stability of Different Convolutional Neural Networks Against Additive Noise},
booktitle={Proceedings of the 12th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications - Volume 5: VISAPP, (VISIGRAPP 2017)},
year={2017},
pages={362-369},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0006200003620369},
isbn={978-989-758-226-4},
}

in EndNote Style

TY - CONF
JO - Proceedings of the 12th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications - Volume 5: VISAPP, (VISIGRAPP 2017)
TI - Studying Stability of Different Convolutional Neural Networks Against Additive Noise
SN - 978-989-758-226-4
AU - Aghdam H.
AU - J. Heravi E.
AU - Puig D.
PY - 2017
SP - 362
EP - 369
DO - 10.5220/0006200003620369