vermamachinelearning / keras-deep-graph-learning Goto Github PK

Why does attention on multi-graph not improving scores ? After 500 epochs scores are lower than without attention.

Hi, Could you please mention what versions of keras and tensorflow is used in this project? I'm trying to import GraphCNN and getting Attribute errors and module errors. I'm trying to run this in Google Colab.

Thank you.

Hi,
Thanks a lot for your effort to write this model.
I want to know which paper you followed for your GraphAttentionCNN model?? I wanted to learn more details about that method.

Thanks
Mohaiminul

How is the data in A_edge_matrices_mutag.csv structured?

This package is good, but is missing a setup.py, and therefore cannot be cloned.

I don't know if I'm taking something wrong, but the very first example of GCN node classification (namely, gcnn_node_classification_example.py) is reaching only 30% accuracy. I have not changed any parameter, just runned the code as it is. Here are the last print lines:

Epoch: 0497 train_acc= 0.3000 test_acc= 0.2970
Epoch: 0498 train_acc= 0.3000 test_acc= 0.2970
Epoch: 0499 train_acc= 0.2929 test_acc= 0.2990

Anyone?

Found this Error after downloading the code and running one of the examples! any help, please.

Got input shape error when trying to do prediction after the model is trained. What should be the input shape? I use model.predict([X, A]).

It seems the code for "gcnn_node_classification_with_edge_features_example.py" is the same as "gcnn_node_classification_example.py". Would you please upload the correct file?
Could you please explain how do we consider edge features in the Adjuscency matrix?