Transcription of Practical Secure Aggregation for Privacy-Preserving ...
{{id}} {{{paragraph}}}
Practical Secure Aggregationfor Privacy-Preserving Machine LearningKeith Bonawitz*, Vladimir Ivanov*, Ben Kreuter*,Antonio Marcedone ,H. Brendan McMahan*, Sarvar Patel*,Daniel Ramage*, Aaron Segal*, and Karn Mountain View, CA 94043 Tech, 2 West Loop Rd., New York, NY 100441 INTRODUCTIONM achine learning models trained on sensitive real-world datapromise improvements to everything from medical screen-ing [48] to disease outbreak discovery [39]. And the wide-spread use of mobile devices means even richer and moresensitive data is becoming available [37].However, large-scale collection of sensitive data entails particularly high-profile example of the consequences ofmishandling sensitive data occurred in 1988, when the videorental history of a nominee for the US Supreme Court waspublished without his consent [4].
withdistinctfieldelementsinF.Giventheseparameters,the scheme consists of two algorithms. The sharing algorithm SS.share(s,t,U) →{(u,s u)} u∈U takes as input a secret s, a set Uof nfield elements representing user IDs, and
Domain:
Source:
Link to this page:
Please notify us if you found a problem with this document:
{{id}} {{{paragraph}}}