Adversarial Autoencoders

TL;DR

The paper introduces adversarial autoencoders, a probabilistic autoencoder framework that leverages GANs to match the posterior distribution with a prior, enabling meaningful sampling and diverse applications.

Contribution

It presents a novel autoencoder model that uses adversarial training to perform variational inference and match the posterior to an arbitrary prior distribution.

Findings

Achieves competitive results in generative modeling

Effective in semi-supervised classification

Useful for disentangling style and content

Abstract

In this paper, we propose the "adversarial autoencoder" (AAE), which is a probabilistic autoencoder that uses the recently proposed generative adversarial networks (GAN) to perform variational inference by matching the aggregated posterior of the hidden code vector of the autoencoder with an arbitrary prior distribution. Matching the aggregated posterior to the prior ensures that generating from any part of prior space results in meaningful samples. As a result, the decoder of the adversarial autoencoder learns a deep generative model that maps the imposed prior to the data distribution. We show how the adversarial autoencoder can be used in applications such as semi-supervised classification, disentangling style and content of images, unsupervised clustering, dimensionality reduction and data visualization. We performed experiments on MNIST, Street View House Numbers and Toronto Face datasets and show that adversarial autoencoders achieve competitive results in generative modeling and semi-supervised classification tasks.