Fabian Jogl presented the paper “The Expressive Power of Path-Based Graph Neural Networks” at the 41st International Conference on Machine Learning (ICML 2024) in Vienna. The work introduces PATH-WL, a general class of color refinement algorithms based on

paths and shortest path distance information, and a corresponding GNN, to systematically analyze the expressive power of path-based graph neural networks. Path-based GNNs are shown to be incomparable to a wide range of expressive GNNs, capable of counting

cycles, and achieving strong results on the challenging family of strongly regular graphs, forming a new hierarchy of highly expressive graph neural networks.