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.
Original language | English |
---|---|
Title of host publication | 12th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications, VISIGRAPP 2017; Porto; Portugal |
Pages | 362-369 |
ISBN (Electronic) | 978-989758226-4 |
DOIs | |
Publication status | Published - 2017 |
Externally published | Yes |
Event | 12th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications: VISAPP 2017 - Porto, Portugal Duration: 27 Feb 2017 → 1 Mar 2017 Conference number: 12 http://www.grapp.visigrapp.org/?y=2017 |
Conference
Conference | 12th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications |
---|---|
Abbreviated title | VISIGRAPP |
Country/Territory | Portugal |
City | Porto |
Period | 27/02/17 → 1/03/17 |
Internet address |