AI-powered
podcast player
Listen to all your favourite podcasts with AI-powered features
The Importance of Locality Laws in Federated Learning
I'm not an expert on federated learning or anything, but I know there are techniques like that that are supposed to allow you to still aggregate data while respecting data privacy. Are those sorts of techniques things that you're able to think about using in this domain? No, that's that's spot on. I think the embedding approach I described earlier is like a first stepping stone on this way to doing full federated learning. And it's an active area that we're still working on and also paying close attention to how these data locality laws are going to evolve so that we can plan the evolution of our own systems.