Building Google Maps' Algorithm & AI Research at Google Deepmind - The What's AI Podcast Episode 17
What's AI Podcast by Louis-François Bouchard
The Importance of Algorithmic Alignment in Graph Machine Learning
This architectural idea can actually give benefits. We trained the model using the principle of algorithmic alignment to simulate the individual steps of a max flow algorithm, which as the name implies, literally computes the flow properties of a particular network. And then we took this pre-trained model and we deployed it inside a state-of-the-art graph machine learning system that predicts these blood vessel properties. That's super interesting. Could you give some not practical, but applied types of data that this can work with? You gave the example for transformers, but where do you come into play with this different architecture? So I can give you a few concrete examples that we published beforehand that show that
Remember Everything You Learn from Podcasts
