papers AI Learner
The Github is limit! Click to go to the new site.

Improved generator objectives for GANs

2016-12-08
Ben Poole, Alexander A. Alemi, Jascha Sohl-Dickstein, Anelia Angelova

Abstract

We present a framework to understand GAN training as alternating density ratio estimation and approximate divergence minimization. This provides an interpretation for the mismatched GAN generator and discriminator objectives often used in practice, and explains the problem of poor sample diversity. We also derive a family of generator objectives that target arbitrary $f$-divergences without minimizing a lower bound, and use them to train generative image models that target either improved sample quality or greater sample diversity.

Abstract (translated by Google)
URL

https://arxiv.org/abs/1612.02780

PDF

https://arxiv.org/pdf/1612.02780


Similar Posts

Comments