N-Gram Graph: Simple Unsupervised Representation for Graphs, with Applications to Molecules

Authors: Shengchao Liu, Mehmet F. Demirel, Yingyu Liang

NeurIPS 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Experiments on 60 tasks from 10 benchmark datasets demonstrate its advantages over both popular graph neural networks and traditional representation methods.
Researcher Affiliation Academia Shengchao Liu, Mehmet Furkan Demirel, Yingyu Liang Department of Computer Sciences, University of Wisconsin-Madison, Madison, WI {shengchao, demirel, yliang}@cs.wisc.edu
Pseudocode Yes Algorithm 1 Vertex Embedding and Algorithm 2 Graph Embedding are provided on page 3.
Open Source Code Yes The code is available at https://github.com/chao1224/n_gram_graph. Baseline implementation follows [21, 44].
Open Datasets Yes Regression datasets: Delaney [18], Malaria [23], CEP [29], QM7 [8], QM8 [43], QM9 [46]. Classification datasets: Tox21 [51], Clin Tox [24, 7], MUV [45], HIV [1].
Dataset Splits Yes All datasets are split into five folds and with cross-validation results reported as follows.
Hardware Specification No The paper mentions 'CPU' and 'GPU' in Table 4 for timing measurements, and acknowledges 'computing resources from the University of Wisconsin-Madison Center for High Throughput Computing', but it does not specify any particular models (e.g., 'NVIDIA A100' or 'Intel Xeon').
Software Dependencies No The paper mentions 'Py Torch Geometric' as part of the baseline implementation in footnote 3 (referencing [21]), but it does not provide specific version numbers for this or any other software dependencies.
Experiment Setup Yes We tune the hyperparameter carefully for all representation and modeling methods. More details about hyperparameters are provided in Section Appendix F. The following subsections display results with the N-gram parameter T = 6 and the embedding dimension r = 100.