GraphRNN: Generating Realistic Graphs with Deep Auto-regressive Models

Authors: Jiaxuan You, Rex Ying, Xiang Ren, William Hamilton, Jure Leskovec

ICML 2018 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Our experiments show that Graph RNN significantly outperforms all baselines, learning to generate diverse graphs that match the structural characteristics of a target set, while also scaling to graphs 50 larger than previous deep models.
Researcher Affiliation Academia 1Department of Computer Science, Stanford University, Stanford, CA, 94305 2Department of Computer Science, University of Southern California, Los Angeles, CA, 90007.
Pseudocode Yes Algorithm 1 Graph RNN inference algorithm
Open Source Code Yes 1The code is available in https://github.com/ snap-stanford/Graph RNN
Open Datasets Yes Community. 500 two-community graphs...generated by the Erd os-R enyi model (E-R) (Erd os & R enyi, 1959)...Protein. 918 protein graphs (Dobson & Doig, 2003)...Ego. 757 3-hop ego networks extracted from the Citeseer network (Sen et al., 2008)
Dataset Splits No We use 80% of the graphs in each dataset for training and test on the rest.
Hardware Specification No The paper does not provide specific hardware details (e.g., GPU/CPU models, memory) used for running the experiments.
Software Dependencies No The paper mentions implementing the model but does not specify any software dependencies with version numbers (e.g., specific deep learning frameworks like PyTorch or TensorFlow versions).
Experiment Setup No The hyperparameter settings for Graph RNN were fixed after development tests on data that was not used in follow-up evaluations (further details in the Appendix).