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

Malicious server (MS) attacks have enabled the scaling of data stealing infederated learning to large batch sizes and secure aggregation, settingspreviously considered private. However, many concerns regarding client-sidedetectability of MS attacks were raised, questioning their practicality oncethey are publicly known. In this work, for the first time, we thoroughly studythe problem of client-side detectability.We demonstrate that most prior MSattacks, which fundamentally rely on one of two key principles, are detectableby principled client-side checks. Further, we formulate desiderata forpractical MS attacks and propose SEER, a novel attack framework that satisfiesall desiderata, while stealing user data from gradients of realistic networks,even for large batch sizes (up to 512 in our experiments) and under secureaggregation. The key insight of SEER is the use of a secret decoder, which isjointly trained with the shared model. Our work represents a promising firststep towards more principled treatment of MS attacks, paving the way forrealistic data stealing that can compromise user privacy in real-worlddeployments.

Hiding in Plain Sight: Disguising Data Stealing Attacks in Federated Learning