Wei Jin

Thanks for reporting the problem. The code at Github has already fixed this issue while we haven't updated it to the pypi source. We will update pip version soon.

Hi, have you tried the script in https://github.com/ChandlerBang/Pro-GNN/blob/master/scripts/meta/cora_meta.sh?

Hi, you may try tune the threshold in GCNJaccard in the range of {0.01, 0.02, 0.03, 0.04, 0.05, 0.1}. This hyper-parameter is important to its performance.

Hey, you may check answer in this issue #12. Feel free to reach out if you have other questions.

The sequential attack method follows the way described in the paper [1]. If you wanna check the original setting in [2], please check [this script](https://github.com/DSE-MSU/DeepRobust/blob/master/examples/graph/test_nettack.py). [1] Robust graph convolutional networks...

Hey, it should be fine to use other GNNs as the surrogate model but you need to assign some new attributes such as ` surrogate.hidden_sizes, surrogate.nfeat, surrogate.nclass`. However, we should...

Hey, you understanding it correct. ProGNN follows [RGCN](https://pengcui.thumedialab.com/papers/RGCN.pdf) to select nodes in the test set whose degrees are larger than 10 and these target nodes are attacked to form one...

你好，这里的脚本已经给出了具体的超参数了，请参考 https://github.com/ChandlerBang/Pro-GNN/tree/master/scripts

In our experiments, we actually consider both transductive (semi-supervised training) and inductive settings.  > in the Cora dataset, with a compression rate of 0.1, the dimension of the adjacency...

- Argument 'hidden' controls the vector dimension of nodes? - Yes - The resulting size of the node vectors is the hidden value (128) or is it adjusted by the...