McGan: Mean and Covariance Feature Matching GAN

Youssef Mroueh, Tom Sercu, Vaibhava Goel
Proceedings of the 34th International Conference on Machine Learning, PMLR 70:2527-2535, 2017.

Abstract

We introduce new families of Integral Probability Metrics (IPM) for training Generative Adversarial Networks (GAN). Our IPMs are based on matching statistics of distributions embedded in a finite dimensional feature space. Mean and covariance feature matching IPMs allow for stable training of GANs, which we will call McGan. McGan minimizes a meaningful loss between distributions.

Cite this Paper

BibTeX
@InProceedings{pmlr-v70-mroueh17a, title = {{M}c{G}an: Mean and Covariance Feature Matching {GAN}}, author = {Youssef Mroueh and Tom Sercu and Vaibhava Goel}, booktitle = {Proceedings of the 34th International Conference on Machine Learning}, pages = {2527--2535}, year = {2017}, editor = {Precup, Doina and Teh, Yee Whye}, volume = {70}, series = {Proceedings of Machine Learning Research}, month = {06--11 Aug}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v70/mroueh17a/mroueh17a.pdf}, url = {https://proceedings.mlr.press/v70/mroueh17a.html}, abstract = {We introduce new families of Integral Probability Metrics (IPM) for training Generative Adversarial Networks (GAN). Our IPMs are based on matching statistics of distributions embedded in a finite dimensional feature space. Mean and covariance feature matching IPMs allow for stable training of GANs, which we will call McGan. McGan minimizes a meaningful loss between distributions.} }
Endnote
%0 Conference Paper %T McGan: Mean and Covariance Feature Matching GAN %A Youssef Mroueh %A Tom Sercu %A Vaibhava Goel %B Proceedings of the 34th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2017 %E Doina Precup %E Yee Whye Teh %F pmlr-v70-mroueh17a %I PMLR %P 2527--2535 %U https://proceedings.mlr.press/v70/mroueh17a.html %V 70 %X We introduce new families of Integral Probability Metrics (IPM) for training Generative Adversarial Networks (GAN). Our IPMs are based on matching statistics of distributions embedded in a finite dimensional feature space. Mean and covariance feature matching IPMs allow for stable training of GANs, which we will call McGan. McGan minimizes a meaningful loss between distributions.
APA
Mroueh, Y., Sercu, T. & Goel, V.. (2017). McGan: Mean and Covariance Feature Matching GAN. Proceedings of the 34th International Conference on Machine Learning, in Proceedings of Machine Learning Research 70:2527-2535 Available from https://proceedings.mlr.press/v70/mroueh17a.html.

Related Material