Optimal Transport Based Generative Autoencoders

TL;DR

This paper introduces two novel autoencoders, AE-OTtrans and AE-OTgen, based on optimal transport, which outperform GANs and other non-adversarial models in image quality and diversity without adversarial training.

Contribution

The paper presents two new generative autoencoders that use optimal transport, avoiding adversarial training and better preserving data manifolds for higher quality images.

Findings

AE-OTtrans and AE-OTgen outperform GANs on MNIST and FashionMNIST.

They achieve state-of-the-art results on MNIST, FashionMNIST, and CelebA datasets.

They produce higher quality and more diverse images than previous models.

Abstract

The field of deep generative modeling is dominated by generative adversarial networks (GANs). However, the training of GANs often lacks stability, fails to converge, and suffers from model collapse. It takes an assortment of tricks to solve these problems, which may be difficult to understand for those seeking to apply generative modeling. Instead, we propose two novel generative autoencoders, AE-OTtrans and AE-OTgen, which rely on optimal transport instead of adversarial training. AE-OTtrans and AEOTgen, unlike VAE and WAE, preserve the manifold of the data; they do not force the latent distribution to match a normal distribution, resulting in greater quality images. AEOTtrans and AE-OTgen also produce images of higher diversity compared to their predecessor, AE-OT. We show that AE-OTtrans and AE-OTgen surpass GANs in the MNIST and FashionMNIST datasets. Furthermore, We show that AE-OTtrans and AE-OTgen do state of the art on the MNIST, FashionMNIST, and CelebA image sets comapred to other non-adversarial generative models.