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

Wireless channels can be inherently privacy-preserving by distorting thereceived signals due to channel noise, and superpositioning multiple signalsover-the-air. By harnessing these natural distortions and superpositions bywireless channels, we propose a novel privacy-preserving machine learning (ML)framework at the network edge, coined over-the-air mixup ML (AirMixML). InAirMixML, multiple workers transmit analog-modulated signals of their privatedata samples to an edge server who trains an ML model using the receivednoisy-and superpositioned samples. AirMixML coincides with model training usingmixup data augmentation achieving comparable accuracy to that with raw datasamples. From a privacy perspective, AirMixML is a differentially private (DP)mechanism limiting the disclosure of each worker's private sample informationat the server, while the worker's transmit power determines the privacydisclosure level. To this end, we develop a fractional channel-inversion powercontrol (PC) method, {\alpha}-Dirichlet mixup PC (DirMix({\alpha})-PC), whereinfor a given global power scaling factor after channel inversion, each worker'slocal power contribution to the superpositioned signal is controlled by theDirichlet dispersion ratio {\alpha}. Mathematically, we derive a closed-formexpression clarifying the relationship between the local and global PC factorsto guarantee a target DP level. By simulations, we provide DirMix({\alpha})-PCdesign guidelines to improve accuracy, privacy, and energy-efficiency. Finally,AirMixML with DirMix({\alpha})-PC is shown to achieve reasonable accuracycompared to a privacy-violating baseline with neither superposition nor PC.

AirMixML: Over-the-Air Data Mixup for Inherently Privacy-Preserving Edge Machine Learning