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.
Youssef Mroueh (IBM T.J Watson Research Center)
Tom Sercu (IBM Research)
Vaibhava Goel (IBM)
Related Events (a corresponding poster, oral, or spotlight)
2017 Talk: McGan: Mean and Covariance Feature Matching GAN »
Mon Aug 7th 03:48 -- 04:06 AM Room Parkside 1