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

Preserving privacy in contemporary NLP models allows us to work withsensitive data, but unfortunately comes at a price. We know that stricterprivacy guarantees in differentially-private stochastic gradient descent(DP-SGD) generally degrade model performance. However, previous research on theefficiency of DP-SGD in NLP is inconclusive or even counter-intuitive. In thisshort paper, we provide an extensive analysis of different privacy preservingstrategies on seven downstream datasets in five different `typical' NLP taskswith varying complexity using modern neural models based on BERT andXtremeDistil architectures. We show that unlike standard non-private approachesto solving NLP tasks, where bigger is usually better, privacy-preservingstrategies do not exhibit a winning pattern, and each task and privacy regimerequires a special treatment to achieve adequate performance.

One size does not fit all: Investigating strategies for differentially-private learning across NLP tasks