vgraphrnn / vgrnn

Variational Graph Recurrent Neural Networks - PyTorch

Variational Graph Recurrent Neural Networks

This is a PyTorch implementation of the VGRNN model as described in our paper:

E. Hajiramezanali*, A. Hasanzadeh*, N. Duffield, K. R. Narayanan, M. Zhou, and X. Qian, Variational Graph Recurrent Neural Networks, Advances in Neural Information Processing Systems (NeurIPS), 2019, *equal contribution

Abstract: Representation learning over graph structured data has been mostly studied in static graph settings while efforts for modeling dynamic graphs are still scant. In this paper, we develop a novel hierarchical variational model that introduces additional latent random variables to jointly model the hidden states of a graph recurrent neural network (GRNN) to capture both topology and node attribute changes in dynamic graphs. We argue that the use of high-level latent random variables in this variational GRNN (VGRNN) can better capture potential variability observed in dynamic graphs as well as the uncertainty of node latent representation. With semi-implicit variational inference developed for this new VGRNN architecture (SI-VGRNN), we show that flexible non-Gaussian latent representations can further help dynamic graph analytic tasks. Our experiments with multiple real-world dynamic graph datasets demonstrate that SI-VGRNN and VGRNN consistently outperform the existing baseline and state-of-the-art methods by a significant margin in dynamic link prediction.

Example: VGRNN learns interpretable latent representations.

Requirements

CUDA==9.0.176
Python==2.7.12
networkx==2.2
scipy==1.1.0
torch==1.0.0      
torch-cluster==1.2.3      
torch-geometric==1.0.2      
torch-scatter==1.1.1      
torch-sparse==0.2.3      
torch-spline-conv==1.0.5      
torchvision==0.2.1

Cite

Please cite our paper if you use this code in your own work:

@inproceedings{hajiramezanali2019variational,
  title={Variational graph recurrent neural networks},
  author={Hajiramezanali, Ehsan and Hasanzadeh, Arman and Narayanan, Krishna and Duffield, Nick and Zhou, Mingyuan and Qian, Xiaoning},
  booktitle={Advances in Neural Information Processing Systems},
  pages={10700--10710},
  year={2019}
}

Please direct your inquiries to [email protected] or [email protected] .

vgrnn's Issues

Failed to load data

Hi,

I'm trying to run your code but always fail to load the data at this line. Please find below the error log:

UnicodeDecodeError: 'ascii' codec can't decode byte 0x96 in position 8: ordinal not in range(128)

Thanks for your help!

Incorrect use of torch.nn.utils.clip_grad_norm

In both training examples clip_grad_norm() is called after opt.step() so it isn't doing anything. It should either be removed or moved before the optimizer updates the model components

Bug in prediction code?

Hello,

I am looking at the VGRNN_prediction.py script and I found what seems to be a bug on this line:

nll_loss += self._nll_bernoulli(dec_t_sl, adj_orig_dense_list[t])

Shouldn't it be:

nll_loss += self._nll_bernoulli(dec_t_sl, adj_orig_dense_list[t + 1])

in order to do link prediction for the graph at time T+1?
Did I misunderstand something?

Thanks

Where is SI-VGRNN? What is the difference beween 'vgrnn_detection' and 'vgrnn_prediction'?

Hi there,

I have some questions about the code. First is where is SIVGRNN? The second one is what is the difference between 'vgrnn_detection' and 'vgrnn_prediction'？

Thanks for your reply!

Pre-processing code for HEP-TH, Cora, and Social Evolution

Hi,

I've been reading your paper and I wanted to play a bit with the datasets that you used for the experiments. I see that you have uploaded the pre-processed data for Facebook, Enron, and Colab, but I couldn't find the other three datasets.
Do you have the code to produce the time series for HEP-TH, Cora, and Social Evolution? Or even the pre-processed data like for the other datasets would be good.

Thanks,
Daniele