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

Data Augmentation (DA) is frequently used to automatically provide additionaltraining data without extra human annotation. However, data augmentation mayintroduce noisy data that impairs training. To guarantee the quality ofaugmented data, existing methods either assume no noise exists in the augmenteddata and adopt consistency training or use simple heuristics such as trainingloss and diversity constraints to filter out ``noisy'' data. However, thosefiltered examples may still contain useful information, and dropping themcompletely causes loss of supervision signals. In this paper, based on theassumption that the original dataset is cleaner than the augmented data, wepropose an on-the-fly denoising technique for data augmentation that learnsfrom soft augmented labels provided by an organic teacher model trained on thecleaner original data. A simple self-regularization module is applied to forcethe model prediction to be consistent across two distinct dropouts to furtherprevent overfitting on noisy labels. Our method can be applied to augmentationtechniques in general and can consistently improve the performance on both textclassification and question-answering tasks.

On-the-fly Denoising for Data Augmentation in Natural Language Understanding