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

Due to limited communication capacities of edge devices, most existingfederated learning (FL) methods randomly select only a subset of devices toparticipate in training for each communication round. Compared with engagingall the available clients, the random-selection mechanism can lead tosignificant performance degradation on non-IID (independent and identicallydistributed) data. In this paper, we show our key observation that theessential reason resulting in such performance degradation is theclass-imbalance of the grouped data from randomly selected clients. Based onour key observation, we design an efficient heterogeneity-aware client samplingmechanism, i.e., Federated Class-balanced Sampling (Fed-CBS), which caneffectively reduce class-imbalance of the group dataset from the intentionallyselected clients. In particular, we propose a measure of class-imbalance andthen employ homomorphic encryption to derive this measure in aprivacy-preserving way. Based on this measure, we also design acomputation-efficient client sampling strategy, such that the actively selectedclients will generate a more class-balanced grouped dataset with theoreticalguarantees. Extensive experimental results demonstrate Fed-CBS outperforms thestatus quo approaches. Furthermore, it achieves comparable or even betterperformance than the ideal setting where all the available clients participatein the FL training.

Fed-CBS: A Heterogeneity-Aware Client Sampling Mechanism for Federated Learning via Class-Imbalance Reduction