- TL;DR: We introduce adversarial training on real samples that does not exist in standard GANs to make discriminator more robust, which can stabilize training, accelerate convergence, and achieve better performance.
- Abstract: Generative adversarial networks have achieved remarkable performance on various tasks but suffer from sensitivity to hyper-parameters, training instability, and mode collapse. We find that this is partly due to gradient given by non-robust discriminator containing non-informative adversarial noise, which can hinder generator from catching the pattern of real samples. Inspired by defense against adversarial samples, we introduce adversarial training of discriminator on real samples that does not exist in classic GANs framework to make adversarial training symmetric, which can balance min-max game and make discriminator more robust. Robust discriminator can give more informative gradient with less adversarial noise, which can stabilize training and accelerate convergence. We validate the proposed method on image generation tasks with varied network architectures quantitatively. Experiments show that training stability, perceptual quality, and diversity of generated samples are consistently improved with small additional training computation cost.
- Keywords: ADVERSARIAL SAMPLES, ADVERSARIAL NETWORKS