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/

Zendy Plus

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

Zendy Tools

Zendy Open

Nowadays, owners and developers of deep learning models must considerstringent privacy-preservation rules of their training data, usuallycrowd-sourced and retaining sensitive information. The most widely adoptedmethod to enforce privacy guarantees of a deep learning model nowadays relieson optimization techniques enforcing differential privacy. According to theliterature, this approach has proven to be a successful defence against severalmodels' privacy attacks, but its downside is a substantial degradation of themodels' performance. In this work, we compare the effectiveness of thedifferentially-private stochastic gradient descent (DP-SGD) algorithm againststandard optimization practices with regularization techniques. We analyze theresulting models' utility, training performance, and the effectiveness ofmembership inference and model inversion attacks against the learned models.Finally, we discuss differential privacy's flaws and limits and empiricallydemonstrate the often superior privacy-preserving properties of dropout andl2-regularization.

On the utility and protection of optimization with differential privacy and classic regularization techniques