XGNN Inference

[voetberg]

Impliment XGNN on NuGraph - include visuals and local inference outputs.

Reference paper - https://dl.acm.org/doi/pdf/10.1145/3394486.3403085

Should be usable with the module from #1

[voetberg]

DIG impliments xgnn and many other methods - Docs here

[voetberg]

Further examination of xgnn reveals from further complexity - The method requires you to train a sub GCN to generate the subgraphs, and that requires some rules for construction of the subgraphs instead of selecting an actual point in the data or producing a mask.

Because gnnexplainer produces basically the same kind of result - I'm shelving this one for now. We can reopen at a later date.

Also a better solution would be to add a few output layers onto the existing network and fine tune the output.