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

In this paper, we propose a variational autoencoder with disentanglementpriors, VAE-DPRIOR, for task-specific natural language generation with none ora handful of task-specific labeled examples. In order to tackle compositionalgeneralization across tasks, our model performs disentangled representationlearning by introducing a conditional prior for the latent content space andanother conditional prior for the latent label space. Both types of priorssatisfy a novel property called $\epsilon$-disentangled. We show bothempirically and theoretically that the novel priors can disentanglerepresentations even without specific regularizations as in the prior work. Thecontent prior enables directly sampling diverse content representations fromthe content space learned from the seen tasks, and fuse them with therepresentations of novel tasks for generating semantically diverse texts in thelow-resource settings. Our extensive experiments demonstrate the superiorperformance of our model over competitive baselines in terms of i) dataaugmentation in continuous zero/few-shot learning, and ii) text style transferin the few-shot setting.

Variational Autoencoder with Disentanglement Priors for Low-Resource Task-Specific Natural Language Generation