Discriminator-Quality Evaluation GAN

Citations Over Time

Abstract

In existing generative adversarial networks (standard GAN and its variants), the discriminator is trained for recognizing the real data as positive while the generated data as negative. This kind of positive-negative classification criterion ignores the fact that the discriminator is a non-objective evaluator, which means that the image quality evaluated by the discriminator may fluctuate during the whole training progress. Considering this fact, we propose a novel GAN framework called Discriminator-Quality Evaluation GAN (DQE-GAN) by using the discriminator outputs to evaluate image quality. By dynamically classifying images into high discriminator-quality and low discriminator-quality samples, every adversarial iteration step can be more reasonable and objective. The convergence of DQE-GAN framework can be theoretically proved. Through extensive experiments, we demonstrate DQE-GANs’ ability of achieving better generated images faster and more stable.

Related Papers

• → GAN-based models and applications(2022)4 cited
• → Differentially private generative decomposed adversarial network for vertically partitioned data sharing(2022)9 cited
• → Student Action Recognition Based on Deep Convolutional Generative Adversarial Network(2020)13 cited
• → DiscrimNet: Semi-Supervised Action Recognition from Videos using\n Generative Adversarial Networks(2018)16 cited
• → Improving Super-Resolution GAN Performance through Discriminator using U-Net Structure and Auxiliary Classifier(2021)