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/

Zendy Plus

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

Zendy Tools

Zendy Open

Federated learning (FL) is a distributed learning paradigm that preservesusers' data privacy while leveraging the entire dataset of all participants. InFL, multiple models are trained independently on the clients and aggregatedcentrally to update a global model in an iterative process. Although thisapproach is excellent at preserving privacy, FL still suffers from qualityissues such as attacks or byzantine faults. Recent attempts have been made toaddress such quality challenges on the robust aggregation techniques for FL.However, the effectiveness of state-of-the-art (SOTA) robust FL techniques isstill unclear and lacks a comprehensive study. Therefore, to better understandthe current quality status and challenges of these SOTA FL techniques in thepresence of attacks and faults, we perform a large-scale empirical study toinvestigate the SOTA FL's quality from multiple angles of attacks, simulatedfaults (via mutation operators), and aggregation (defense) methods. Inparticular, we study FL's performance on the image classification tasks and useDNNs as our model type. Furthermore, we perform our study on two generic imagedatasets and one real-world federated medical image dataset. We alsoinvestigate the effect of the proportion of affected clients and the datasetdistribution factors on the robustness of FL. After a large-scale analysis with496 configurations, we find that most mutators on each user have a negligibleeffect on the final model in the generic datasets, and only one of them iseffective in the medical dataset. Furthermore, we show that model poisoningattacks are more effective than data poisoning attacks. Moreover, choosing themost robust FL aggregator depends on the attacks and datasets. Finally, weillustrate that a simple ensemble of aggregators achieves a more robustsolution than any single aggregator and is the best choice in 75% of the cases.

Towards Understanding Quality Challenges of the Federated Learning for Neural Networks: A First Look from the Lens of Robustness