Thomas Kipf

Duplicate of https://github.com/tkipf/gcn/issues/113 On Wed, May 8, 2019 at 8:04 PM aubutt wrote: > Hi there is on things I want to ask . It can be a dumb question...

Have a look at https://github.com/tkipf/pygcn/blob/master/pygcn/utils.py#L15 for a more instructive example for how to create your own data loader. On Wed, May 29, 2019 at 11:41 PM tiredDi wrote: > sorry...

X denote the node features. On Mon, Apr 1, 2019 at 1:05 PM Junlei Yang wrote: > Hello, thanks for your excellent work. I am new in this area and...

That highly depends on the dataset, but you could still get reasonable predictions. On Thu 4. Apr 2019 at 08:41 Wangzhen wrote: > Hello again， > I want to know...

Scikit learn has some convenience functions for creating confusion matrices: http://scikit-learn.org/stable/modules/generated/sklearn.metrics.confusion_matrix.html Just provide the predicted classes and the true labels. You might need to perform an argmax or max operation...

Our R-GCN model can handle this setting out-of-the-box, as it does not matter how many edges of different type there are between two nodes. If you would like to predict...

As far as I know these train/val/test splits should be disjunct. Why do you think that there is an overlap?

Thanks for your comment. This is indeed a (small) difference and it should not affect performance in any significant way. I expect that one could show that both formulations are...

Yes, I believe either the frame_skip between the individual actions (or for how many frames an action is applied) is not fully deterministic in the gym environment that our codebase...

Thank you for your question! We use our own implementation for this baseline, following the architecture as described in Appendix D.5 of our paper. Essentially, the CNN encoder architecture is...