Generative adversarial interpolative autoencoding: adversarial training on latent space interpolations encourage convex latent distributions

TL;DR

This paper introduces a novel neural network architecture combining autoencoders and GANs to promote convex latent distributions, enabling realistic image interpolations and high-quality sample generation.

Contribution

It proposes a new adversarial training method on latent space interpolations using an autoencoder as both generator and discriminator.

Findings

Produces non-blurry, high-quality samples

Interpolations stay within the real image distribution

Maintains realistic features in generated images

Abstract

We present a neural network architecture based upon the Autoencoder (AE) and Generative Adversarial Network (GAN) that promotes a convex latent distribution by training adversarially on latent space interpolations. By using an AE as both the generator and discriminator of a GAN, we pass a pixel-wise error function across the discriminator, yielding an AE which produces non-blurry samples that match both high- and low-level features of the original images. Interpolations between images in this space remain within the latent-space distribution of real images as trained by the discriminator, and therfore preserve realistic resemblances to the network inputs. Code available at https://github.com/timsainb/GAIA