kl_fake: KL WGAN loss on fake examples

In RGAN: Generative Adversarial Nets (GAN) in R

KL WGAN loss on fake examples

Description

Utility function for the kl WGAN loss for fake examples as described in https://arxiv.org/abs/1910.09779 and implemented in python in https://github.com/ermongroup/f-wgan.

Usage

kl_fake(dis_fake)

Arguments

dis_fake

Discriminator scores on fake examples ($D(G(z))$)

Value

The loss

RGAN documentation built on March 30, 2022, 1:07 a.m.