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) enables multiple clients to train a model withoutcompromising sensitive data. The decentralized nature of FL makes itsusceptible to adversarial attacks, especially backdoor insertion duringtraining. Recently, the edge-case backdoor attack employing the tail of thedata distribution has been proposed as a powerful one, raising questions aboutthe shortfall in current defenses' robustness guarantees. Specifically, mostexisting defenses cannot eliminate edge-case backdoor attacks or suffer from atrade-off between backdoor-defending effectiveness and overall performance onthe primary task. To tackle this challenge, we propose FedGrad, a novelbackdoor-resistant defense for FL that is resistant to cutting-edge backdoorattacks, including the edge-case attack, and performs effectively underheterogeneous client data and a large number of compromised clients. FedGrad isdesigned as a two-layer filtering mechanism that thoroughly analyzes theultimate layer's gradient to identify suspicious local updates and remove themfrom the aggregation process. We evaluate FedGrad under different attackscenarios and show that it significantly outperforms state-of-the-art defensemechanisms. Notably, FedGrad can almost 100% correctly detect the maliciousparticipants, thus providing a significant reduction in the backdoor effect(e.g., backdoor accuracy is less than 8%) while not reducing the main accuracyon the primary task.

FedGrad: Mitigating Backdoor Attacks in Federated Learning Through Local Ultimate Gradients Inspection