Awesome
keras-aae
Reproduces Adversarial Autoencoder architecture from Makhzani, Alireza, et al. "Adversarial autoencoders." arXiv preprint arXiv:1511.05644 (2015) with Keras.
Summary
The Adversarial Autoencoder behaves similarly to Variational Autoencoders, forcing the latent space of an autoencoder to follow a predefined prior. In the case of the Adversarial Autoencoder, this latent space can be defined arbitrarily and easily sampled and fed into the Discriminator in the network.
<div> <img src="https://raw.githubusercontent.com/greentfrapp/keras-aae/master/images/aae_latent.png" alt="Latent space from Adversarial Autoencoder" width="whatever" height="300px" style="display: inline-block;"> <img src="https://raw.githubusercontent.com/greentfrapp/keras-aae/master/images/regular_latent.png" alt="Latent space from regular Autoencoder" width="whatever" height="300px" style="display: inline-block;"> </div>The left image shows the latent space of an unseen MNIST test set after training with an Adversarial Autoencoder for 50 epochs, which follows a 2D Gaussian prior. Contrast this with the latent space of the regular Autoencoder trained under the same conditions, with a far more irregular latent distribution.
Instructions
To train a model just run
$ python keras-aae.py --train
For more parameters, run with --help flag.
For comparison with a regular autoencoder, run
$ python regular-ae.py --train --noadversarial