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 the context of neural machine translation, data augmentation (DA)techniques may be used for generating additional training samples when theavailable parallel data are scarce. Many DA approaches aim at expanding thesupport of the empirical data distribution by generating new sentence pairsthat contain infrequent words, thus making it closer to the true datadistribution of parallel sentences. In this paper, we propose to follow acompletely different approach and present a multi-task DA approach in which wegenerate new sentence pairs with transformations, such as reversing the orderof the target sentence, which produce unfluent target sentences. Duringtraining, these augmented sentences are used as auxiliary tasks in a multi-taskframework with the aim of providing new contexts where the target prefix is notinformative enough to predict the next word. This strengthens the encoder andforces the decoder to pay more attention to the source representations of theencoder. Experiments carried out on six low-resource translation tasks showconsistent improvements over the baseline and over DA methods aiming atextending the support of the empirical data distribution. The systems trainedwith our approach rely more on the source tokens, are more robust againstdomain shift and suffer less hallucinations.

Rethinking Data Augmentation for Low-Resource Neural Machine Translation: A Multi-Task Learning Approach