Federated learning

A couple of years ago, I was the lead author on a report about Federated learning.

Federated learning is a family of algorithms for doing distributed machine learning without moving the training data between machines. This can reduce the bandwidth and power requirements, and it helps to preserve the privacy of the data.

The report goes into all this in much more detail. It will take you about an hour to read.

It used to be behind a paywall, available only to clients of Fast Forward Labs (once a startup, now a part of Cloudera). But it was recently made public, so I hope you’ll check it out!

If you’d rather watch a talk about federated learning, then I gave one at Strange Loop in 2019 (slides).

If you’d rather listen to a podcast, then I spoke with the wonderful Jeffrey Meyerson on Software Engineering Daily.

And if you’d rather play a slightly weird federated learning game then check out Turbofan Tycoon, the prototype I built with Grant Custer!