Discriminator optimal transport
This work addresses image generation quality for GAN users, offering incremental improvements through a novel theoretical insight applied to existing models.
The paper tackles the problem of improving generative adversarial networks (GANs) by showing that discriminator optimization approximates optimal transport between generated and target distributions, and proposes a discriminator optimal transport (DOT) scheme that improves inception score and FID on datasets like CIFAR-10, STL-10, and ImageNet.
Within a broad class of generative adversarial networks, we show that discriminator optimization process increases a lower bound of the dual cost function for the Wasserstein distance between the target distribution $p$ and the generator distribution $p_G$. It implies that the trained discriminator can approximate optimal transport (OT) from $p_G$ to $p$.Based on some experiments and a bit of OT theory, we propose a discriminator optimal transport (DOT) scheme to improve generated images. We show that it improves inception score and FID calculated by un-conditional GAN trained by CIFAR-10, STL-10 and a public pre-trained model of conditional GAN by ImageNet.