Thomas Kipf

You can simply set `y_train=y_val=y_test` and also provide the full adjacency matrix and the full feature matrix (i.e., an identity matrix) during training and testing. It suffices to simply provide...

You can make predictions on new nodes if their feature vectors come from a similar underlying distribution. If you take unique one-hot vectors as feature vectors (i.e., an identity matrix),...

Yes and yes! > On 5 Nov 2018, at 13:24, Riccardo Bucco wrote: > > What's the dimension of the three layers? > > input --> 34, output --> out1...

Same On Mon 5. Nov 2018 at 14:53 Riccardo Bucco wrote: > what about the out1 dimension? I can't find its value anywhere > > — > You are receiving...

We provide a more comprehensive data loader function in this repository: https://github.com/tkipf/pygcn Hope this helps. On Fri 8. Jun 2018 at 10:15 Rajeshneti wrote: > Hi, > I need to...

Yes, this is intentional (see paper). On Tue 10. Sep 2019 at 07:47 ZhanqiLiu wrote: > [image: image] > > > — > You are receiving this because you are...

Looks good to me on first glance. Make sure that the function `tf.nn.sigmoid_cross_entropy_with_logits` receives logits and not outputs from a sigmoid. On Tue, May 28, 2019 at 4:48 PM Chengmeng94...

You should threshold the output of the sigmoid (`preds`) before comparing it to the binary labels. On Wed, May 29, 2019 at 4:48 AM Chengmeng94 wrote: > It seems that...

Yes this feature was intended. Unfortunately the API of tensorflow changed in the meantime which gives you this error I suppose. You might want to try feeding the adjacency matrices...

Each ‘support’ is an adjacency matrix, so you can see it as multiple graphs, yes. On Tue 11. Sep 2018 at 09:03 Tong Guo wrote: > multiple "support" means multiple...