Generative Adversarial Networks (GANs) are a class of generative algorithms
that have been shown to produce state-of-the art samples, especially in the
domain of image creation. The fundamental principle of GANs is to approximate
the unknown distribution of a given data set by optimizing an objective
function through an adversarial game between a family of generators and a
family of discriminators. In this paper, we offer a better theoretical
understanding of GANs by analyzing some of their mathematical and statistical
properties. We study the deep connection between the adversarial principle
underlying GANs and the Jensen-Shannon divergence, together with some
optimality characteristics of the problem. An analysis of the role of the
discriminator family via approximation arguments is also provided. In addition,
taking a statistical point of view, we study the large sample properties of the
estimated distribution and prove in particular a central limit theorem. Some of
our results are illustrated with simulated examples.
%0 Generic
%1 biau2018theoretical
%A Biau, G.
%A Cadre, B.
%A Sangnier, M.
%A Tanielian, U.
%D 2018
%K 2018 GAN arxiv deep-learning paper
%T Some Theoretical Properties of GANs
%U http://arxiv.org/abs/1803.07819
%X Generative Adversarial Networks (GANs) are a class of generative algorithms
that have been shown to produce state-of-the art samples, especially in the
domain of image creation. The fundamental principle of GANs is to approximate
the unknown distribution of a given data set by optimizing an objective
function through an adversarial game between a family of generators and a
family of discriminators. In this paper, we offer a better theoretical
understanding of GANs by analyzing some of their mathematical and statistical
properties. We study the deep connection between the adversarial principle
underlying GANs and the Jensen-Shannon divergence, together with some
optimality characteristics of the problem. An analysis of the role of the
discriminator family via approximation arguments is also provided. In addition,
taking a statistical point of view, we study the large sample properties of the
estimated distribution and prove in particular a central limit theorem. Some of
our results are illustrated with simulated examples.
@misc{biau2018theoretical,
abstract = {Generative Adversarial Networks (GANs) are a class of generative algorithms
that have been shown to produce state-of-the art samples, especially in the
domain of image creation. The fundamental principle of GANs is to approximate
the unknown distribution of a given data set by optimizing an objective
function through an adversarial game between a family of generators and a
family of discriminators. In this paper, we offer a better theoretical
understanding of GANs by analyzing some of their mathematical and statistical
properties. We study the deep connection between the adversarial principle
underlying GANs and the Jensen-Shannon divergence, together with some
optimality characteristics of the problem. An analysis of the role of the
discriminator family via approximation arguments is also provided. In addition,
taking a statistical point of view, we study the large sample properties of the
estimated distribution and prove in particular a central limit theorem. Some of
our results are illustrated with simulated examples.},
added-at = {2018-03-22T16:47:11.000+0100},
author = {Biau, G. and Cadre, B. and Sangnier, M. and Tanielian, U.},
biburl = {https://www.bibsonomy.org/bibtex/202a1cbdb828ae05ab93678859ab5cca2/achakraborty},
description = {[1803.07819] Some Theoretical Properties of GANs},
interhash = {2f300accc9475371698b9402d46bd0c5},
intrahash = {02a1cbdb828ae05ab93678859ab5cca2},
keywords = {2018 GAN arxiv deep-learning paper},
note = {cite arxiv:1803.07819},
timestamp = {2018-03-22T16:47:11.000+0100},
title = {Some Theoretical Properties of GANs},
url = {http://arxiv.org/abs/1803.07819},
year = 2018
}