Import Error

[vvshyer]

After installing pynauty, run the first cell in receptive_field_maker_example.ipynb returned importError:

ImportError: dlopen(/Users/sunhuayan/Library/Python/3.6/lib/python/site-packages/pynauty/nautywrap.cpython-36m-darwin.so, 2): Symbol not found: _dispatch_graph Referenced from: /Users/sunhuayan/Library/Python/3.6/lib/python/site-packages/pynauty/nautywrap.cpython-36m-darwin.so Expected in: flat namespace in /Users/sunhuayan/Library/Python/3.6/lib/python/site-packages/pynauty/nautywrap.cpython-36m-darwin.so

Could you please tell me how to fix this? Many THX~

[tvayer]

It's working for me in Mac OS 10.12 and Unbutu 16.04 with both python 3.5 and 3.6. What is your OS ?

Maybe you could check : https://stackoverflow.com/questions/35006614/what-does-symbol-not-found-expected-in-flat-namespace-actually-mean https://stackoverflow.com/questions/32277351/what-does-expected-in-flat-namespace-mean

[vvshyer]

Mac OS 10.14 with python 3.6

[tvayer]

I updated Nauty, uninstall pynauty and rebuild it with the new version and tell me if you still get the error

[vvshyer]

It works~ Thank you very much.

[tvayer]

Great !

[tvayer]

Just download the datasets here : https://ls11-www.cs.tu-dortmund.de/staff/morris/graphkerneldatasets. They are in the smae format used in data_loader.py, just wirte a custom function in the same way

Le lun. 5 nov. 2018 à 07:45, 孙华岩 notifications@github.com a écrit :

Hi, I'm sorry to bother you again. I want to reproduce the experiment of this paper. These are data sets (here is the link: http://www.mit.edu/~pinary/kdd/) in the paper I've found. Could you please tell me how to convert these data sets into the format of the data set you provided? Thank you so much.

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/tvayer/PSCN/issues/2#issuecomment-435770668, or mute the thread https://github.com/notifications/unsubscribe-auth/AQYWDdZ7zPQ2FVO7qAsnpWxf7zGWl1tTks5ur96IgaJpZM4YBWZW .

[vvshyer]

I found that the accuracy of y_test in train_example.ipynb is only 30+. Why it's so bad? Is it possible to make it better?

[tvayer]

Yes there was an error in preds, the output was a 2 -dim array, i added an update (I just used ravel for preds)

[vvshyer]

Thanks for updating! I still confused about how to make receptive field for edges? I mean the whole paper is talking about nodes. I am thinking about what if I want to learn CNNs for both nodes and edges? Have you every think about this? Thank you very much again~