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

Can a pre-trained generator be adapted to the hybrid of multiple targetdomains and generate images with integrated attributes of them? In this work,we introduce a new task -- Few-shot Hybrid Domain Adaptation (HDA). Given asource generator and several target domains, HDA aims to acquire an adaptedgenerator that preserves the integrated attributes of all target domains,without overriding the source domain's characteristics. Compared with DomainAdaptation (DA), HDA offers greater flexibility and versatility to adaptgenerators to more composite and expansive domains. Simultaneously, HDA alsopresents more challenges than DA as we have access only to images fromindividual target domains and lack authentic images from the hybrid domain. Toaddress this issue, we introduce a discriminator-free framework that directlyencodes different domains' images into well-separable subspaces. To achieveHDA, we propose a novel directional subspace loss comprised of a distance lossand a direction loss. Concretely, the distance loss blends the attributes ofall target domains by reducing the distances from generated images to alltarget subspaces. The direction loss preserves the characteristics from thesource domain by guiding the adaptation along the perpendicular to subspaces.Experiments show that our method can obtain numerous domain-specific attributesin a single adapted generator, which surpasses the baseline methods in semanticsimilarity, image fidelity, and cross-domain consistency.

Few-shot Hybrid Domain Adaptation of Image Generators