search

divelab / DIG

A library for graph deep learning research
https://diveintographs.readthedocs.io/
GNU General Public License v3.0
1.84k stars 281 forks source link

GNNExplainer: optimization benchmark #111

Closed alirezadizaji closed 2 years ago

alirezadizaji commented 2 years ago

as it seems, an explanation output is generated by the GNNExplainer per each label exists in the dataset. isn't better to only define the benchmark based on the original model prediction?

Oceanusity commented 2 years ago

Hello, thank you for your issue. Since the explanations for all the labels are provided, you can select the one corresponding to the original model prediction as the final explanation.