In this paper we present emph{system}, a decentralized, edge-based
framework that supports heterogeneous privacy policies for federated learning.
We evaluate our system on three use cases that train models with sensitive user
data collected by mobile phones — predictive text, image classification, and
notification engagement prediction — on a Raspberry~Pi edge device. We find
that system is able to perform accurate model training and inference within
reasonable resource and time budgets while also enforcing heterogeneous privacy
policies.

Go to Source of this post
Author Of this post: <a href="http://arxiv.org/find/cs/1/au:+Katevas_K/0/1/0/all/0/1">Kleomenis Katevas</a>, <a href="http://arxiv.org/find/cs/1/au:+Bagdasaryan_E/0/1/0/all/0/1">Eugene Bagdasaryan</a>, <a href="http://arxiv.org/find/cs/1/au:+Waterman_J/0/1/0/all/0/1">Jason Waterman</a>, <a href="http://arxiv.org/find/cs/1/au:+Safadieh_M/0/1/0/all/0/1">Mohamad Mounir Safadieh</a>, <a href="http://arxiv.org/find/cs/1/au:+Birrell_E/0/1/0/all/0/1">Eleanor Birrell</a>, <a href="http://arxiv.org/find/cs/1/au:+Haddadi_H/0/1/0/all/0/1">Hamed Haddadi</a>, <a href="http://arxiv.org/find/cs/1/au:+Estrin_D/0/1/0/all/0/1">Deborah Estrin</a>

By admin