Question about edge pruning

[Chenhui1016]

Hi,

Thanks for the great work! I have a question for the code

GNNGuard/defense/gcn.py

Line 184 in 33f5390

sim[sim<0.1] = 0

, where it drops the edge based on the similarity score in Equation 3 of the paper. However, the paper only mentions that dropping the edge is based on the Equation 5, not Equation 3. Could you kindly explain why do we add this line in the code? Or something I'm missing?

Thanks!

[yzfxmu]

Hi,

Thanks for the great work! I have a question for the code

GNNGuard/defense/gcn.py

Line 184 in 33f5390

sim[sim<0.1] = 0

, where it drops the edge based on the similarity score in Equation 3 of the paper. However, the paper only mentions that dropping the edge is based on the Equation 5, not Equation 3. Could you kindly explain why do we add this line in the code? Or something I'm missing?
Thanks!

I have the same question.

[Jieerbobo]

Hi,

Thanks for the great work! I have a question for the code

GNNGuard/defense/gcn.py

Line 184 in 33f5390

sim[sim<0.1] = 0

, where it drops the edge based on the similarity score in Equation 3 of the paper. However, the paper only mentions that dropping the edge is based on the Equation 5, not Equation 3. Could you kindly explain why do we add this line in the code? Or something I'm missing?
Thanks!

As indicated in "Are Defenses for Graph Neural Networks Robust?", It is vital to do this adjustment to remove edges with too dissimilar node embeddings (which would, as experiments show, substantially hinder attacking).
Paper link: https://www.cs.cit.tum.de/daml/are-gnn-defenses-robust/