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 has emerged as a popular technique for distributingmachine learning (ML) model training across the wireless edge. In this paper,we propose two timescale hybrid federated learning (TT-HF), asemi-decentralized learning architecture that combines the conventionaldevice-to-server communication paradigm for federated learning withdevice-to-device (D2D) communications for model training. In TT-HF, during eachglobal aggregation interval, devices (i) perform multiple stochastic gradientdescent iterations on their individual datasets, and (ii) aperiodically engagein consensus procedure of their model parameters through cooperative,distributed D2D communications within local clusters. With a new generaldefinition of gradient diversity, we formally study the convergence behavior ofTT-HF, resulting in new convergence bounds for distributed ML. We leverage ourconvergence bounds to develop an adaptive control algorithm that tunes the stepsize, D2D communication rounds, and global aggregation period of TT-HF overtime to target a sublinear convergence rate of O(1/t) while minimizing networkresource utilization. Our subsequent experiments demonstrate that TT-HFsignificantly outperforms the current art in federated learning in terms ofmodel accuracy and/or network energy consumption in different scenarios wherelocal device datasets exhibit statistical heterogeneity. Finally, our numericalevaluations demonstrate robustness against outages caused by fading channels,as well favorable performance with non-convex loss functions.

Semi-Decentralized Federated Learning with Cooperative D2D Local Model Aggregations