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

Despite advances in multilingual neural machine translation (MNMT), we arguethat there are still two major challenges in this area: data imbalance andrepresentation degeneration. The data imbalance problem refers to the imbalancein the amount of parallel corpora for all language pairs, especially forlong-tail languages (i.e., very low-resource languages). The representationdegeneration problem refers to the problem of encoded tokens tending to appearonly in a small subspace of the full space available to the MNMT model. Tosolve these two issues, we propose Bi-ACL, a framework that uses onlytarget-side monolingual data and a bilingual dictionary to improve theperformance of the MNMT model. We define two modules, named bidirectionalautoencoder and bidirectional contrastive learning, which we combine with anonline constrained beam search and a curriculum learning sampling strategy.Extensive experiments show that our proposed method is more effective both inlong-tail languages and in high-resource languages. We also demonstrate thatour approach is capable of transferring knowledge between domains and languagesin zero-shot scenarios.

Mitigating Data Imbalance and Representation Degeneration in Multilingual Machine Translation