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

Federated Learning (FL) allows multiple participating clients to trainmachine learning models collaboratively by keeping their datasets local andonly exchanging the gradient or model updates with a coordinating server.Existing FL protocols were shown to be vulnerable to attacks that aim tocompromise data privacy and/or model robustness. Recently proposed defensesfocused on ensuring either privacy or robustness, but not both. In this paper,we focus on simultaneously achieving differential privacy (DP) and Byzantinerobustness for cross-silo FL, based on the idea of learning from history. Therobustness is achieved via client momentum, which averages the updates of eachclient over time, thus reduces the variance of the honest clients and exposesthe small malicious perturbations of Byzantine clients that are undetectable ina single round but accumulate over time. In our initial solution DP-BREM, theDP property is achieved via adding noise to the aggregated momentum, and weaccount for the privacy cost from the momentum, which is different from theconventional DP-SGD that accounts for the privacy cost from gradient. SinceDP-BREM assumes a trusted server (who can obtain clients' local models orupdates), we further develop the final solution called DP-BREM+, which achievesthe same DP and robustness properties as DP-BREM without a trusted server byutilizing secure aggregation techniques, where DP noise is securely and jointlygenerated by the clients. Our theoretical analysis on the convergence rate andexperimental results under different DP guarantees and attack settingsdemonstrate that our proposed protocols achieve better privacy-utility tradeoffand stronger Byzantine robustness than several baseline methods.

DP-BREM: Differentially-Private and Byzantine-Robust Federated Learning with Client Momentum