Just created a list of papers about the expressiveness of graph neural networks (GNNs). Hope this is helpful for people who are interested in this area. Please let me know or submit a pull request if I missed any important related work.
Conversation
Replying to
Since you study the field, do you have a recommendation on which GNN algorithms are practical? I don't see any other significant work from GNN, except the protein folding.
1
Replying to
This is really a big topic. We have seen many simple and scalable GNNs are applied in the industry. The papers we collected here are mainly about improving the expressiveness of GNNs. For sure, they are not ready to take care of industry-scale graphs.
1
