/complex-kaf-demo

Demo of the complex KAF neural network

Primary LanguageJupyter NotebookApache License 2.0Apache-2.0

Content

This is a demo of the complex-valued KAF neural network proposed in this paper: https://arxiv.org/abs/1802.08026

We train the network on the Fashion MNIST dataset, after extracting the frequency components from every image.

Prerequisites

The code needs the Autograd library: https://github.com/HIPS/autograd

Code for cloning the Fashion MNIST repository (https://github.com/zalandoresearch/fashion-mnist) is provided in the notebook.

Execution

Simply runs the steps in the notebook to (1) load data, (2) define the models, and (3) train it.

Contact

For any additional information, please contact me at simone [dot] scardapane [at] uniroma1 [dot] it.