2022 Talk: Beyond Homophily In Graph Neural Networks (GNNs)
Please enjoy this talk “Beyond Homophily in Graph Neural Networks (GNNs),” from the Knowledge Graph Conference 2024 given by Danai Koutra, Professor at the University of Michigan.
Talk description:
In this talk, the University of Michigan’s Professor Danai Koutra presents her group’s research on identifying effective graph neural network (GNN) designs and introduces a new architectures that improves performance in heterophilous networks. Professor Koutra also discusses connections between heterophily and other well-known challenges of GNNs, including oversmoothing and robustness.
Do you enjoy talks like this one? Subscribe to the KGC newsletter and stay up to date with everything happening in the world of knowledge graphs, AI, and semantic tech.
***
