English

GAN and VAE from an Optimal Transport Point of View

Machine Learning 2017-06-07 v1

Abstract

This short article revisits some of the ideas introduced in arXiv:1701.07875 and arXiv:1705.07642 in a simple setup. This sheds some lights on the connexions between Variational Autoencoders (VAE), Generative Adversarial Networks (GAN) and Minimum Kantorovitch Estimators (MKE).

Keywords

Cite

@article{arxiv.1706.01807,
  title  = {GAN and VAE from an Optimal Transport Point of View},
  author = {Aude Genevay and Gabriel Peyré and Marco Cuturi},
  journal= {arXiv preprint arXiv:1706.01807},
  year   = {2017}
}