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

Neural machine translation usually adopts autoregressive models and suffersfrom exposure bias as well as the consequent error propagation problem. Manyprevious works have discussed the relationship between error propagation andthe \emph{accuracy drop} (i.e., the left part of the translated sentence isoften better than its right part in left-to-right decoding models) problem. Inthis paper, we conduct a series of analyses to deeply understand this problemand get several interesting findings. (1) The role of error propagation onaccuracy drop is overstated in the literature, although it indeed contributesto the accuracy drop problem. (2) Characteristics of a language play a moreimportant role in causing the accuracy drop: the left part of the translationresult in a right-branching language (e.g., English) is more likely to be moreaccurate than its right part, while the right part is more accurate for aleft-branching language (e.g., Japanese). Our discoveries are confirmed ondifferent model structures including Transformer and RNN, and in other sequencegeneration tasks such as text summarization.

Beyond Error Propagation in Neural Machine Translation: Characteristics of Language Also Matter