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

Data poisoning and backdoor attacks manipulate training data to inducesecurity breaches in a victim model. These attacks can be provably deflectedusing differentially private (DP) training methods, although this comes with asharp decrease in model performance. The InstaHide method has recently beenproposed as an alternative to DP training that leverages supposed privacyproperties of the mixup augmentation, although without rigorous guarantees. Inthis work, we show that strong data augmentations, such as mixup and randomadditive noise, nullify poison attacks while enduring only a small accuracytrade-off. To explain these finding, we propose a training method,DP-InstaHide, which combines the mixup regularizer with additive noise. Arigorous analysis of DP-InstaHide shows that mixup does indeed have privacyadvantages, and that training with k-way mixup provably yields at least k timesstronger DP guarantees than a naive DP mechanism. Because mixup (as opposed tonoise) is beneficial to model performance, DP-InstaHide provides a mechanismfor achieving stronger empirical performance against poisoning attacks thanother known DP methods.

DP-InstaHide: Provably Defusing Poisoning and Backdoor Attacks with Differentially Private Data Augmentations