Timezone: »

 
Poster
Understanding and Stabilizing GANs' Training Dynamics Using Control Theory
Kun Xu · Chongxuan Li · Jun Zhu · Bo Zhang

Thu Jul 16 07:00 AM -- 07:45 AM & Thu Jul 16 08:00 PM -- 08:45 PM (PDT) @ None #None
in Poster Session 42 »

Generative adversarial networks (GANs) are effective in generating realistic images but the training is often unstable. There are existing efforts that model the training dynamics of GANs in the parameter space but the analysis cannot directly motivate practically effective stabilizing methods. To this end, we present a conceptually novel perspective from control theory to directly model the dynamics of GANs in the frequency domain and provide simple yet effective methods to stabilize GAN's training. We first analyze the training dynamic of a prototypical Dirac GAN and adopt the widely-used closed-loop control (CLC) to improve its stability. We then extend CLC to stabilize the training dynamic of normal GANs, which can be implemented as an L2 regularizer on the output of the discriminator. Empirical results show that our method can effectively stabilize the training and obtain state-of-the-art performance on data generation tasks.

Keywords: Deep Generative Models · Generative Adversarial Networks · Deep Learning - General

Author Information

Taufik Xu (Tsinghua University)
Chongxuan Li (Tsinghua University)
Jun Zhu (Tsinghua University)
Bo Zhang (Tsinghua University)

More from the Same Authors