EUVIMEDEUVIMED entdecken
Crossref

ADGNN: Towards Scalable GNN Training with Aggregation-Difference Aware Sampling

Zhen Song, Yu Gu, Tianyi Li, Qing Sun, Yanfeng Zhang, Christian S. Jensen, Ge Yu

Proceedings of the ACM on Management of Data · 08.12.2023 · 1 (4) · 1-26

DOI: 10.1145/3626716 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Proceedings of the ACM on Management of Data, online erschienen am 12.12.2023. Verfasst von Zhen Song und 6 weiteren Autor:innen. Die Arbeit wurde bisher 9-mal zitiert (Crossref) und nennt 53 Literaturangaben. Gefördert von: the National Natural Science Foundation of China, the Fundamental Research Funds for the Central Universities.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Distributed computing is promising to enable large-scale graph neural network (GNN) model training. However, care is needed to avoid excessive computational and communication overheads. Sampling is promising in terms of enabling scalability, and sampling techniques have been proposed to reduce training costs. However, online sampling introduces large overheads, and while offline sampling that is done only once can eliminate such overheads, it instead introduces information loss and accuracy degradation. Thus, existing sampling techniques are unable to improve simultaneously both efficiency and accuracy, particularly at low sampling rates. We develop a distributed system, ADGNN, for full-batch based GNN training that adopts a hybrid sampling architecture to enable a trade-off between efficiency and accuracy. Specifically, ADGNN employs sampling result reuse techniques to reduce the cost associated with sampling and thus improve training efficiency. To alleviate accuracy degradation, we introduce a new metric,Aggregation Difference (AD), that quantifies the gap between sampled and full neighbor set aggregation. We present so-called AD-Sampling that aims to minimize the Aggregation Difference with an adaptive sampling frequency tuner. Finally, ADGNN employs anAD -importance-based sampling technique for remote neighbors to further reduce communication costs. Experiments on five real datasets show that ADGNN is able to outperform the state-of-the-art by up to nearly 9 times in terms of efficiency, while achieving comparable accuracy to the non-sampling methods.

Förderung

  • the National Natural Science Foundation of China
  • the Fundamental Research Funds for the Central Universities

Zitieren

APA

Song, Z., Gu, Y., Li, T., Sun, Q., Zhang, Y., Jensen, C. S., & Yu, G. (2023). ADGNN: Towards Scalable GNN Training with Aggregation-Difference Aware Sampling. Proceedings of the ACM on Management of Data, 1(4), 1-26. https://doi.org/10.1145/3626716

Vancouver

Song Z, Gu Y, Li T, Sun Q, Zhang Y, Jensen CS, et al. ADGNN: Towards Scalable GNN Training with Aggregation-Difference Aware Sampling. Proceedings of the ACM on Management of Data. 2023;1(4):1-26. doi:10.1145/3626716

BibTeX
@article{Song2023,
  author = {Song, Zhen and Gu, Yu and Li, Tianyi and Sun, Qing and Zhang, Yanfeng and Jensen, Christian S. and Yu, Ge},
  title = {ADGNN: Towards Scalable GNN Training with Aggregation-Difference Aware Sampling},
  journal = {Proceedings of the ACM on Management of Data},
  year = {2023},
  volume = {1},
  number = {4},
  pages = {1--26},
  doi = {10.1145/3626716}
}
Kostenlos starten →