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.
|Title of host publication||12th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications, VISIGRAPP 2017; Porto; Portugal|
|Publication status||Published - 2017|
|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
|Conference||12th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications|
|Period||27/02/17 → 1/03/17|