English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Conditional Generative Adversarial Networks (cGANs) are implicit generativemodels which allow to sample from class-conditional distributions. ExistingcGANs are based on a wide range of different discriminator designs and trainingobjectives. One popular design in earlier works is to include a classifierduring training with the assumption that good classifiers can help eliminatesamples generated with wrong classes. Nevertheless, including classifiers incGANs often comes with a side effect of only generating easy-to-classifysamples. Recently, some representative cGANs avoid the shortcoming and reachstate-of-the-art performance without having classifiers. Somehow it remainsunanswered whether the classifiers can be resurrected to design better cGANs.In this work, we demonstrate that classifiers can be properly leveraged toimprove cGANs. We start by using the decomposition of the joint probabilitydistribution to connect the goals of cGANs and classification as a unifiedframework. The framework, along with a classic energy model to parameterizedistributions, justifies the use of classifiers for cGANs in a principledmanner. It explains several popular cGAN variants, such as ACGAN, ProjGAN, andContraGAN, as special cases with different levels of approximations, whichprovides a unified view and brings new insights to understanding cGANs.Experimental results demonstrate that the design inspired by the proposedframework outperforms state-of-the-art cGANs on multiple benchmark datasets,especially on the most challenging ImageNet. The code is available athttps://github.com/sian-chen/PyTorch-ECGAN.

A Unified View of cGANs with and without Classifiers