Fact · Voice privacy and security
McMahan and colleagues proposed federated learning, an approach that leaves the training data distributed on mobile devices and learns a shared model by aggregating locally computed updates.
Checked against its source on October 17, 2026
In the source's own words
“leaves the training data distributed on the mobile devices, and learns a shared model by aggregating locally-computed updates”
Communication-Efficient Learning of Deep Networks from Decentralized Data, H. Brendan McMahan et al., Proceedings of the 20th International Conference on Artificial Intelligence and Statistics (AISTATS), JMLR Workshop and Conference Proceedings volume 54, 2017, Abstract.
Open the original- What it measures
- The paper names the approach federated learning. The abstract says its empirical evaluation covers five different model architectures and four datasets.
- What it does not show
- The paper is about the training method and its communication cost. It does not claim that model updates reveal nothing about the data, and it does not measure a commercial product.