Reducing Noise in GAN Training with Variance Reduced Extragradient
This addresses convergence issues in GAN training for machine learning practitioners, offering an incremental improvement over existing optimization methods.
The paper tackles the problem of stochastic gradient noise preventing convergence in GAN training by introducing a novel variance-reduced extragradient algorithm, which improves convergence rates and yields more stable training on standard datasets while being computationally cheaper than batch methods.
We study the effect of the stochastic gradient noise on the training of generative adversarial networks (GANs) and show that it can prevent the convergence of standard game optimization methods, while the batch version converges. We address this issue with a novel stochastic variance-reduced extragradient (SVRE) optimization algorithm, which for a large class of games improves upon the previous convergence rates proposed in the literature. We observe empirically that SVRE performs similarly to a batch method on MNIST while being computationally cheaper, and that SVRE yields more stable GAN training on standard datasets.