search

lihan97 / KPGT

codes for KPGT (Knowledge-guided Pre-training of Graph Transformer)
Apache License 2.0
84 stars 14 forks source link

About the selection of the K node #9

Closed Fireflower-March closed 3 months ago

Fireflower-March commented 4 months ago

"Are the masked nodes and the additional knowledge referring to the same portion or just masked in the same proportion?"

lihan97 commented 4 months ago

Thank you for your interest in our work!

During the pre-training process, we mask all the input features for 50% of the nodes in each graph. However, this masking process does not apply to the K node. We specifically mask 50% of its input features for each K node in individual graphs.