Stabilizing GAN Training with Multiple Random Projections
arXiv: Learning, Volume abs/1705.07831, 2018.
We proposed a new framework to training Generative adversarial networks for high-dimensional outputs
Training generative adversarial networks is unstable in high-dimensions as the true data distribution tends to be concentrated in a small fraction of the ambient space. The discriminator is then quickly able to classify nearly all generated samples as fake, leaving the generator without meaningful gradients and causing it to deteriorate a...More
PPT (Upload PPT)