You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I am recently using the GNNexplainer in geometric.explain to analyze the feature importance of a trained GAT model, and having some technical questions regarding the interpretation I can derive from the results.
When I choose "phenomena" and "attributes", to find the overall importance of each feature, should I sum the node score mask over all the nodes?
After getting the importance score for each feature, how should one determine whether a feature is important or not? Is there a general rule to follow to characterize a feature as unimportant for the overall performance?
Is there any other physical metrics which we can convert the importance scores to? (Such as influence on the loss function values)
reacted with thumbs up emoji reacted with thumbs down emoji reacted with laugh emoji reacted with hooray emoji reacted with confused emoji reacted with heart emoji reacted with rocket emoji reacted with eyes emoji
-
Hi PyG team and community,
I am recently using the GNNexplainer in geometric.explain to analyze the feature importance of a trained GAT model, and having some technical questions regarding the interpretation I can derive from the results.
Thank you for your help in advance!
Beta Was this translation helpful? Give feedback.
All reactions