ICLR2020 Spotlight: Rethinking GANs & Realness 🌟

We generalize the standard GAN to a new perspective by treating realness as a random variable that can be estimated from multiple angles (referred as RealnessGAN). While RealnessGAN shares similar theoretical guarantees with the standard GAN, it provides more insights on adversarial learning. We show that RealnessGAN provides stronger guidance for the generator, achieving improvements on both synthetic and real-world datasets. It also enables the basic DCGAN architecture to generate realistic images at 1024x1024 resolution when trained from scratch.

Project page: https://github.com/kam1107/RealnessGAN
Paper: https://arxiv.org/abs/2002.05512