Stabilizing Training of Generative Adversarial Networks through Regularization
This solves the instability problem in GAN training for researchers and practitioners, making GANs more reliable as building blocks in deep learning.
The paper tackles the fragility of GAN training by addressing dimensional mismatch between model and data distributions, proposing a low-cost regularization method that stabilizes training across architectures on benchmark image tasks.
Deep generative models based on Generative Adversarial Networks (GANs) have demonstrated impressive sample quality but in order to work they require a careful choice of architecture, parameter initialization, and selection of hyper-parameters. This fragility is in part due to a dimensional mismatch or non-overlapping support between the model distribution and the data distribution, causing their density ratio and the associated f-divergence to be undefined. We overcome this fundamental limitation and propose a new regularization approach with low computational cost that yields a stable GAN training procedure. We demonstrate the effectiveness of this regularizer across several architectures trained on common benchmark image generation tasks. Our regularization turns GAN models into reliable building blocks for deep learning.