Linking Generative Adversarial Learning and Binary Classification

2017/09/05 by Balsubramani, Akshay
#Artificial Intelligence (cs.AI) #FOS: Computer and information sciences #Machine Learning (cs.LG) #Machine Learning (stat.ML)

paper · doi:10.48550/arxiv.1709.01509

Abstract

In this note, we point out a basic link between generative adversarial (GA) training and binary classification -- any powerful discriminator essentially computes an (f-)divergence between real and generated samples. The result, repeatedly re-derived in decision theory, has implications for GA Networks (GANs), providing an alternative perspective on training f-GANs by designing the discriminator loss function.

Related

• New Losses for Generative Adversarial Learning
• On the Discrimination-Generalization Tradeoff in GANs
• Generalization Error of GAN from the Discriminator's Perspective
• Learning GANs and Ensembles Using Discrepancy
• Many Paths to Equilibrium: GANs Do Not Need to Decrease a Divergence At Every Step