search

HKUDS / GraphEdit

"GraphEdit: Large Language Models for Graph Structure Learning"
https://arxiv.org/abs/2402.15183
Apache License 2.0
120 stars 10 forks source link

question about adversarial robustness experiment #3

Closed zhongjian-zhang closed 7 months ago

zhongjian-zhang commented 7 months ago

Hi, I have a question on "Model Robustness Study against Noise".

  1. Which attack method is used in the experiment?
  2. Did the attack occur during the training phase or the testing phase?

Looking forward your reply! Best wish!

LarFii commented 7 months ago

Thanks for your interest. For the questions:

  1. We use random noise in the experiment by injecting noisy edges into the original graph structure at different rates.
  2. After training the GraphEdit, we inject noise and use the trained GraphEdit to denoise before the GNN training.