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

We show that aggregated model updates in federated learning may be insecure.An untrusted central server may disaggregate user updates from sums of updatesacross participants given repeated observations, enabling the server to recoverprivileged information about individual users' private training data viatraditional gradient inference attacks. Our method revolves aroundreconstructing participant information (e.g: which rounds of training usersparticipated in) from aggregated model updates by leveraging summaryinformation from device analytics commonly used to monitor, debug, and managefederated learning systems. Our attack is parallelizable and we successfullydisaggregate user updates on settings with up to thousands of participants. Wequantitatively and qualitatively demonstrate significant improvements in thecapability of various inference attacks on the disaggregated updates. Ourattack enables the attribution of learned properties to individual users,violating anonymity, and shows that a determined central server may underminethe secure aggregation protocol to break individual users' data privacy infederated learning.

Gradient Disaggregation: Breaking Privacy in Federated Learning by Reconstructing the User Participant Matrix