Please do first (on your own): <a href="https://pytorch.org/tutorials/beginner/blitz/cifar10_tutorial.html">https://pytorch.org/tutorials/beginner/blitz/cifar10_tutorial.html</a>
<br><br>
Then, get <a href="invertible.ipynb">invertible.ipynb</a> and <a href="scattering.ipynb">scattering.ipynb</a>.

<br>
<br>
If you're fast enough, try <a href="https://pytorch.org/tutorials/advanced/neural_style_tutorial.html">https://pytorch.org/tutorials/advanced/neural_style_tutorial.html</a> and change the layers used for style transfer.

<br><br><br><br>

        kernel = self.conv.weight.data<br>
        kernel_fft = fft(kernel,dim=2)<br>
        abs_kernel_fft = torch.abs(kernel_fft)<br>
        renorm_abs_kernel_fft=abs_kernel_fft<br>
        renorm_abs_kernel_fft[abs_kernel_fft<0.9] = 1<br>
        renorm_abs_kernel_fft[abs_kernel_fft>=0.9] *= 1/0.9<br>
        kernel_ = ifft(kernel_fft/renorm_abs_kernel_fft,dim=2)<br>
        self.conv.weight.data.copy_(kernel_.real)<br>