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 Tools annual

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 Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

Free-rider attacks against federated learning consist in dissimulatingparticipation to the federated learning process with the goal of obtaining thefinal aggregated model without actually contributing with any data. This kindof attacks is critical in sensitive applications of federated learning, wheredata is scarce and the model has high commercial value. We introduce here thefirst theoretical and experimental analysis of free-rider attacks on federatedlearning schemes based on iterative parameters aggregation, such as FedAvg orFedProx, and provide formal guarantees for these attacks to converge to theaggregated models of the fair participants. We first show that astraightforward implementation of this attack can be simply achieved by notupdating the local parameters during the iterative federated optimization. Asthis attack can be detected by adopting simple countermeasures at the serverlevel, we subsequently study more complex disguising schemes based onstochastic updates of the free-rider parameters. We demonstrate the proposedstrategies on a number of experimental scenarios, in both iid and non-iidsettings. We conclude by providing recommendations to avoid free-rider attacksin real world applications of federated learning, especially in sensitivedomains where security of data and models is critical.

Free-rider Attacks on Model Aggregation in Federated Learning