Thomas Kipf

Weighting should in principle be fine too, but training dynamics can certainly be affected by whether you choose a weighting strategy or negative sampling. I would always prefer negative sampling...

We are planning to release the code for our Graph Convolutional Matrix Completion project later this month. This is essentially a multi-relational extension of the code from this repository. On...

Yes, have a look here: https://arxiv.org/abs/1609.02907 On Mon 3. Dec 2018 at 06:09 Long Wu wrote: > Hi @tkipf : > I have question about norm in train.py > https://github.com/tkipf/gae/blob/master/gae/train.py...

I did a very small scale grid search around typical values for learning rate, dropout and hidden layer on a validation set. On Sat 15. Dec 2018 at 15:53 Sheikh...

Looks like you have a `nan` or `inf` value somewhere in the model, e.g. in the variable `sample_weight`. You should be able to find the problem with a debugger.

Looks like the loss might become inf or nan at some point, then you get nan-valued gradients. This might come from the KL term in the VAE loss. You can...

Looks like the variance (`tf.exp(self.z_log_std)`) is diverging for some reason. Not sure why. You can try our follow-up model, which should be more stable: https://nicola-decao.github.io/s-vae

Have a look at the output of the `load_data` function:) On Sun 26. May 2019 at 20:21 Feng.Zhang wrote: > In order to use your own data, you have to...

Maybe your classes are imbalanced? On Sun 7. Jul 2019 at 08:15 Saed Rezayi wrote: > I'm using your code on a different dataset (SNOMED), so I use my own...

What I meant is that accuracy in link prediction can be misleading if the classes (link vs no connection) in your evaluation set are imbalanced. On Sun 7. Jul 2019...