...still limited in their expressive power. This means that there are some tasks that GCNs are not able to learn as well as other models, such as deep neural networks. For example, GCNs are not able to learn long-range dependencies between nodes in a graph, which can be a problem for tasks such...