Fisher GAN
This addresses training stability issues for researchers and practitioners using GANs, representing an incremental improvement over existing methods.
The paper tackles the problem of stable training in Generative Adversarial Networks (GANs) by introducing Fisher GAN, which uses a data-dependent constraint on second-order moments, resulting in stable and time-efficient training without compromising critic capacity or needing weight clipping.
Generative Adversarial Networks (GANs) are powerful models for learning complex distributions. Stable training of GANs has been addressed in many recent works which explore different metrics between distributions. In this paper we introduce Fisher GAN which fits within the Integral Probability Metrics (IPM) framework for training GANs. Fisher GAN defines a critic with a data dependent constraint on its second order moments. We show in this paper that Fisher GAN allows for stable and time efficient training that does not compromise the capacity of the critic, and does not need data independent constraints such as weight clipping. We analyze our Fisher IPM theoretically and provide an algorithm based on Augmented Lagrangian for Fisher GAN. We validate our claims on both image sample generation and semi-supervised classification using Fisher GAN.