Towards Understanding the Dynamics of Generative Adversarial Networks

نویسندگان

  • Jerry Li
  • Aleksander Madry
  • John Peebles
  • Ludwig Schmidt
چکیده

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 problematic convergence behaviors (e.g., vanishing gradient, mode collapse, diverging or oscillatory behavior) and allows us to still 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 steps 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.

برای دانلود رایگان متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

منابع مشابه

Automatic Colorization of Grayscale Images Using Generative Adversarial Networks

Automatic colorization of gray scale images poses a unique challenge in Information Retrieval. The goal of this field is to colorize images which have lost some color channels (such as the RGB channels or the AB channels in the LAB color space) while only having the brightness channel available, which is usually the case in a vast array of old photos and portraits. Having the ability to coloriz...

متن کامل

Improvement of generative adversarial networks for automatic text-to-image generation

This research is related to the use of deep learning tools and image processing technology in the automatic generation of images from text. Previous researches have used one sentence to produce images. In this research, a memory-based hierarchical model is presented that uses three different descriptions that are presented in the form of sentences to produce and improve the image. The proposed ...

متن کامل

Towards Principled Methods for Training Generative Adversarial Networks

The goal of this paper is not to introduce a single algorithm or method, but to make theoretical steps towards fully understanding the training dynamics of generative adversarial networks. In order to substantiate our theoretical analysis, we perform targeted experiments to verify our assumptions, illustrate our claims, and quantify the phenomena. This paper is divided into three sections. The ...

متن کامل

Interaction Matters: A Note on Non-asymptotic Local Convergence of Generative Adversarial Networks

Motivated by the pursuit of a systematic computational and algorithmic understanding of Generative Adversarial Networks (GANs), we present a simple yet unified non-asymptotic local convergence theory for smooth two-player games, which subsumes several discrete-time gradientbased saddle point dynamics. The analysis reveals the surprising nature of the off-diagonal interaction term as both a bles...

متن کامل

Mode Regularized Generative Adversarial Networks

Although Generative Adversarial Networks achieve state-of-the-art results on a variety of generative tasks, they are regarded as highly unstable and prone to miss modes. We argue that these bad behaviors of GANs are due to the very particular functional shape of the trained discriminators in high dimensional spaces, which can easily make training stuck or push probability mass in the wrong dire...

متن کامل

ذخیره در منابع من


  با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید

عنوان ژورنال:
  • CoRR

دوره abs/1706.09884  شماره 

صفحات  -

تاریخ انتشار 2017