This repository provides code for reproducing the figures in the paper:

``Deep Decoder: Concise Image Representations from Untrained Non-convolutional Networks'', by Reinhard Heckel and Paul Hand. Contact: [email protected]

The paper is available online [here].

The deep decoder is a simple image generating deep neural network. The network is untrained, non-convolutional, and under-parameterized, i.e., it generates images from few paramters. The deep decoder enables image compression and solving invere problems:

The code is written in python and relies on pytorch. The following libraries are required:

• python 3
• pytorch
• numpy
• skimage
• matplotlib
• scikit-image
• jupyter

The libraries can be installed via:

conda install jupyter

@article{heckel_deep_2018,
    author    = {Reinhard Heckel and Paul Hand},
    title     = {Deep Decoder: Concise Image Representations from Untrained Non-convolutional Networks},
    journal   = {International Conference on Learning Representations},
    year      = {2019}
}

All files are provided under the terms of the Apache License, Version 2.0.