You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
Ali Amiri a14ad2caa8 fix bug in evaluation 1 year ago
GraphCompletion a 1 year ago
baselines as 1 year ago
dataset a 1 year ago
eval a 1 year ago
main_baselines a 1 year ago
temp a 1 year ago
LICENSE a 1 year ago
README.md a 1 year ago
__init__.py a 1 year ago
analysis.py a 1 year ago
args.py a 1 year ago
create_graphs.py a 1 year ago
data.py a 1 year ago
environment.yml a 1 year ago
evaluate.py a 1 year ago
evaluate2.py a 1 year ago
main.py a 1 year ago
main_DeepGMG.py fix bug in evaluation 1 year ago
model.py a 1 year ago
plot.py a 1 year ago
requirements.txt a 1 year ago
test_MMD.py a 1 year ago
train.py a 1 year ago
utils.py a 1 year ago

README.md

GraphRNN: Generating Realistic Graphs with Deep Auto-regressive Model

This repository is the official PyTorch implementation of GraphRNN, a graph generative model using auto-regressive model.

Jiaxuan You*, Rex Ying*, Xiang Ren, William L. Hamilton, Jure Leskovec, GraphRNN: Generating Realistic Graphs with Deep Auto-regressive Model (ICML 2018)

Installation

Install PyTorch following the instuctions on the official website. The code has been tested over PyTorch 0.2.0 and 0.4.0 versions.

conda install pytorch torchvision cuda90 -c pytorch

Then install the other dependencies.

pip install -r requirements.txt

Test run

python main.py

Code description

For the GraphRNN model: main.py is the main executable file, and specific arguments are set in args.py. train.py includes training iterations and calls model.py and data.py create_graphs.py is where we prepare target graph datasets.

For baseline models:

  • B-A and E-R models are implemented in baselines/baseline_simple.py.
  • Kronecker graph model is implemented in the SNAP software, which can be found in https://github.com/snap-stanford/snap/tree/master/examples/krongen (for generating Kronecker graphs), and https://github.com/snap-stanford/snap/tree/master/examples/kronfit (for learning parameters for the model).
  • MMSB is implemented using the EDWARD library (http://edwardlib.org/), and is located in baselines.
  • We implemented the DeepGMG model based on the instructions of their paper in main_DeepGMG.py.
  • We implemented the GraphVAE model based on the instructions of their paper in baselines/graphvae.

Parameter setting: To adjust the hyper-parameter and input arguments to the model, modify the fields of args.py accordingly. For example, args.cuda controls which GPU is used to train the model, and args.graph_type specifies which dataset is used to train the generative model. See the documentation in args.py for more detailed descriptions of all fields.

Outputs

There are several different types of outputs, each saved into a different directory under a path prefix. The path prefix is set at args.dir_input. Suppose that this field is set to ./:

  • ./graphs contains the pickle files of training, test and generated graphs. Each contains a list of networkx object.
  • ./eval_results contains the evaluation of MMD scores in txt format.
  • ./model_save stores the model checkpoints
  • ./nll saves the log-likelihood for generated graphs as sequences.
  • ./figures is used to save visualizations (see Visualization of graphs section).

Evaluation

The evaluation is done in evaluate.py, where user can choose which settings to evaluate. To evaluate how close the generated graphs are to the ground truth set, we use MMD (maximum mean discrepancy) to calculate the divergence between two sets of distributions related to the ground truth and generated graphs. Three types of distributions are chosen: degree distribution, clustering coefficient distribution. Both of which are implemented in eval/stats.py, using multiprocessing python module. One can easily extend the evaluation to compute MMD for other distribution of graphs.

We also compute the orbit counts for each graph, represented as a high-dimensional data point. We then compute the MMD between the two sets of sampled points using ORCA (see http://www.biolab.si/supp/orca/orca.html) at eval/orca. One first needs to compile ORCA by

g++ -O2 -std=c++11 -o orca orca.cpp` 

in directory eval/orca. (the binary file already in repo works in Ubuntu).

To evaluate, run

python evaluate.py

Arguments specific to evaluation is specified in class evaluate.Args_evaluate. Note that the field Args_evaluate.dataset_name_all must only contain datasets that are already trained, by setting args.graph_type to each of the datasets and running python main.py.

Visualization of graphs

The training, testing and generated graphs are saved at ‘graphs/’. One can visualize the generated graph using the function utils.load_graph_list, which loads the list of graphs from the pickle file, and util.draw_graph_list, which plots the graph using networkx.

Misc

Jesse Bettencourt and Harris Chan have made a great slide introducing GraphRNN in Prof. David Duvenaud’s seminar course Learning Discrete Latent Structure.