AI-powered
podcast player
Listen to all your favourite podcasts with AI-powered features
Message Passing in Neural Networks
Zack: The way message passing works in neural networks actually fails in certain cases. So they came up with this idea, why don't we create a policy to essentially decompose a agraph into lots of other grafts? Zack: And that policy could be, ido, get me all the minimum spanning trees of this graft. That would be a dumb thing to do, cause i think they said there's expedentially many of them. But something like that, and thenyo, you stack all o those adjacency matra season, and then you work on that.