RankGAN: A Maximum Margin Ranking GAN for Generating Faces
This work addresses the challenge of enhancing GAN training for face image generation, though it appears incremental as it builds on existing GAN frameworks with a new loss and training strategy.
The authors tackled the problem of improving generative adversarial networks (GANs) for face generation by introducing a stage-wise learning paradigm called RankGAN, which uses a margin-based ranking loss to progressively strengthen the discriminator and generator, resulting in visual and quantitative improvements over methods like WGAN and LSGAN on the CelebA dataset.
We present a new stage-wise learning paradigm for training generative adversarial networks (GANs). The goal of our work is to progressively strengthen the discriminator and thus, the generators, with each subsequent stage without changing the network architecture. We call this proposed method the RankGAN. We first propose a margin-based loss for the GAN discriminator. We then extend it to a margin-based ranking loss to train the multiple stages of RankGAN. We focus on face images from the CelebA dataset in our work and show visual as well as quantitative improvements in face generation and completion tasks over other GAN approaches, including WGAN and LSGAN.