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 Tools annual

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 Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

Simultaneous machine translation (SiMT) generates translation while readingthe whole source sentence. However, existing SiMT models are typically trainedusing the same reference disregarding the varying amounts of available sourceinformation at different latency. Training the model with ground-truth at lowlatency may introduce forced anticipations, whereas utilizing referenceconsistent with the source word order at high latency results in performancedegradation. Consequently, it is crucial to train the SiMT model withappropriate reference that avoids forced anticipations during training whilemaintaining high quality. In this paper, we propose a novel method thatprovides tailored reference for the SiMT models trained at different latency byrephrasing the ground-truth. Specifically, we introduce the tailor, induced byreinforcement learning, to modify ground-truth to the tailored reference. TheSiMT model is trained with the tailored reference and jointly optimized withthe tailor to enhance performance. Importantly, our method is applicable to awide range of current SiMT approaches. Experiments on three translation tasksdemonstrate that our method achieves state-of-the-art performance in both fixedand adaptive policies.

Simultaneous Machine Translation with Tailored Reference