Timezone: »
Despite advances in the field of Graph Neural Networks (GNNs), only a small number (~5) of datasets are currently used to evaluate new models. This continued reliance on a handful of datasets provides minimal insight into the performance differences between models, and is especially challenging for industrial practitioners who have datasets which are very different from academic benchmarks.In this work we introduce GraphWorld, a novel methodology and system for benchmarking GNN models on an arbitrarily-large population of synthetic graphs for any conceivable GNN task. GraphWorld allows a user to efficiently generate a \emph{world} with millions of statistically diverse datasets. It is accessible, scalable, and easy to use. GraphWorld can be run on a single machine without specialized hardware, or it can be easily scaled up to run on arbitrary clusters or cloud frameworks. Using GraphWorld, a user has fine-grained control over graph generator parameters, and can benchmark arbitrary GNN models.We present insights from GraphWorld experiments on the performance of thirteen GNN models and baselines over millions of benchmark datasets. We show that GraphWorld efficiently explores regions of benchmark dataset space uncovered by standard benchmarks, revealing comparisons between models that have not been historically obtainable. Using GraphWorld, we also are able to study in-detail the relationship between graph properties and task performance metrics, which is nearly impossible with the classic collection of real-world benchmarks.
Author Information
John Palowitch (Google)
Anton Tsitsulin (Google)
Bryan Perozzi (Google Research)
Brandon Mayer (Google)
More from the Same Authors
-
2022 : Differentially Private Graph Learning via Sensitivity-Bounded Personalized PageRank »
Alessandro Epasto · Vahab Mirrokni · Bryan Perozzi · Anton Tsitsulin · Peilin Zhong -
2023 Poster: Learning Large Graph Property Prediction via Graph Segment Training »
Kaidi Cao · Phitchaya Phothilimtha · Sami Abu-El-Haija · Dustin Zelle · Yanqi Zhou · Charith Mendis · Jure Leskovec · Bryan Perozzi -
2022 Spotlight: Machine Learning on Graphs: A Model and Comprehensive Taxonomy »
Ines Chami · Sami Abu-El-Haija · Bryan Perozzi · Christopher Ré · Kevin Murphy -
2022 Poster: Differentially Private Graph Learning via Sensitivity-Bounded Personalized PageRank »
Alessandro Epasto · Vahab Mirrokni · Bryan Perozzi · Anton Tsitsulin · Peilin Zhong -
2022 Poster: Zero-shot Transfer Learning within a Heterogeneous Graph via Knowledge Transfer Networks »
Minji Yoon · John Palowitch · Dustin Zelle · Ziniu Hu · Ruslan Salakhutdinov · Bryan Perozzi -
2022 Poster: Machine Learning on Graphs: A Model and Comprehensive Taxonomy »
Ines Chami · Sami Abu-El-Haija · Bryan Perozzi · Christopher Ré · Kevin Murphy -
2022 : Closing »
Anton Tsitsulin -
2022 Expo Workshop: Graph Neural Networks in Tensorflow: A Practical Guide »
Bryan Perozzi · Sami A Abu-El-Haija · Neslihan Bulut · Brandon Mayer -
2022 : Welcome »
Bryan Perozzi -
2021 Poster: Shift-Robust GNNs: Overcoming the Limitations of Localized Graph Training data »
Qi Zhu · Natalia Ponomareva · Jiawei Han · Bryan Perozzi -
2020 : Debiasing GNNs »
John Palowitch -
2020 : GNNs and Graph Embeddings »
Bryan Perozzi -
2020 Expo Workshop: Mining and Learning with Graphs at Scale »
Vahab Mirrokni · Bryan Perozzi · Jakub Lacki · Jonathan Halcrow · Jaqui C Herman