I want to create a framework that allows GDL to be applied to time-varying graphs. I came up with the Erdos-renyi model as an example of a time-varying graphs.
GDL for graphs takes node information as input and takes correspondence with correct data as accuracy. However, how should I deal with time-varying data, even random, and ground-truth data for such graphs? Or is there other better way? And is it nonsense to use pseudo-coordinate as input, as it refers to the traditional approach to time-invariant graphs? Also, an application of time-varying graphs has been anomaly detection in financial networks. How does this work specifically? Also, please let me know if there are other application examples.