install the enviroment failed about graphcl

[yyou1996]

Hi @Austinzhenghua,

Thanks for your interest. For semi_TU experiments, if you fail to work with torch_geometric==1.4.0 which is out-of-dated, you can take a try an upgraded version https://github.com/Shen-Lab/GraphCL_Automated/tree/master/semisupervised_TU that should work on torch_geometric >= 1.6.0.

[Austinzhenghua]

Hi, I have a question about the dataset. If I have my own data but not graph data. Can I use KNN to construct some r ego network to instead the TU dataset?  Thank you so much! hua zheng ***@***.*** 签名由 网易邮箱大师 定制 On 06/29/2021 21:41, Yuning You wrote: Hi @Austinzhenghua, Thanks for your interest. For semi_TU experiments, if you fail to work with torch_geometric==1.4.0 which is out-of-dated, you can take a try an upgraded version https://github.com/Shen-Lab/GraphCL_Automated/tree/master/semisupervised_TU that should work on torch_geometric >= 1.6.0. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or unsubscribe.

[yyou1996]

Yes, you can construct ur dataset by KNN.

[Austinzhenghua]

I use KNN for each data point to construct N subgraph or just use one time to construct a graph? hua zheng ***@***.*** 签名由 网易邮箱大师 定制 On 06/30/2021 02:07, Yuning You wrote: Yes, you can construct ur dataset by KNN. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or unsubscribe.

[yyou1996]

Looks like either work for me.

[Austinzhenghua]

So I should construct the graph data same format like the TU dataset? the raw data have many documents as far as I know. Thank you for your kind help! hua zheng ***@***.*** 签名由 网易邮箱大师 定制 On 06/30/2021 09:28, Yuning You wrote: Looks like either work for me. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or unsubscribe.

[yyou1996]

Yes replace tu_dataset is one way to plugin ur own data.