Generative Adversarial Networks (GANs) have recently been proposed as a promising avenue towards learning generative models with deep neural networks. While GANs have demonstrated state-of-the-art performance on multiple vision tasks, their learning dynamics are not yet well understood, both in theory and in practice. To address this issue, we take a first step towards a rigorous study of GAN dynamics. We propose a simple model that exhibits several of the common problematic convergence behaviors (e.g., vanishing gradient, mode collapse, diverging or oscillatory behavior) and still allows us to establish the first convergence bounds for parametric GAN dynamics. We find an interesting dichotomy: a GAN with an optimal discriminator provably converges, while a first order approximation of the discriminator leads to unstable GAN dynamics and mode collapse. Our model and analysis point to a specific challenge in practical GAN training that we call discriminator collapse.

本文研究生成对抗网络中的学习动态，通过对简单但富有挑战性收敛行为(如梯度消失、模式崩溃、发散或振荡)的参数模型进行理论分析和实践研究，发现使用一阶判别器步骤是导致GAN训练具有挑战性的因素之一。

关于GAN动态中一阶逼近的限制