amin-salehi / gate Goto Github PK

Thank you so much for your kindness.
Now i build a Graph by myself with feature.I want to know how i can make the graph like ind.cora.graph,ind.cora.x,ind.cora.y and so on.
it‘s there some code，so that i can change my graph like this easily.

Originally posted by @hrbwzq in #4 (comment)

How can i use this with my own dataset.
How can i change my graph into this kind of format.
thank you so much.

Hi, I wonder if the training process can be generalized to the minibatch setting when we consider the task of multiple graphs ？
Thank you very much.

Hey Amin,

This paper introduces a very novel idea to learn node representations in an unsupervised way. However, I was wondering if jointly training the autoencoder and classifier by adding the classifier loss to the optimization objective would result in an improvement in performance. This ideology was inspired by the paper titled Semi-Supervised autoencoder. Since, we only use 20 labels per class, I believe this methodology should lead to an improvement in classification performance. Your thoughts on the same would be appreciated.

Thanks

Can you please elaborate in detail that how can we generate the test.index file, while using the code with image dataset?
Please elaborate on the procedure to generate the test.index file.

we are facing a problem with line 87 of the inductive_process.py file, as we only focus on inductive_process this time. Can you please help us?

Thanks in advance...

Could you please tell me that how I can use image dataset.

thank you so much.

Hello,

Thanks for such nice work !! Can you please let me know how can I setup and install GATE on my PC?

Thanks...