跳到主要导航 跳到搜索 跳到主要内容

Brief industry paper: Optimizing memory efficiency of graph neural networks on edge computing platforms

科研成果: 书/报告/会议事项章节会议稿件同行评审

摘要

Graph neural networks (GNN) have achieved state-of-the-art performance on various industrial tasks. However, the poor efficiency of GNN inference and frequent Out-of-Memory (OOM) problem limit the successful application of GNN on edge computing platforms. To tackle these problems, a feature decomposition approach is proposed for memory efficiency optimization of GNN inference. The proposed approach could achieve outstanding optimization on various GNN models, covering a wide range of datasets, which speeds up the inference by up to 3×. Furthermore, the proposed feature decomposition could significantly reduce the peak memory usage (up to 5× in memory efficiency improvement) and mitigate OOM problems during GNN inference.

源语言英语
主期刊名Proceedings - 2021 IEEE 27th Real-Time and Embedded Technology and Applications Symposium, RTAS 2021
出版商Institute of Electrical and Electronics Engineers Inc.
445-448
页数4
ISBN(电子版)9781665403863
DOI
出版状态已出版 - 5月 2021
活动27th IEEE Real-Time and Embedded Technology and Applications Symposium, RTAS 2021 - Virtual, Online
期限: 18 5月 202121 5月 2021

出版系列

姓名Proceedings of the IEEE Real-Time and Embedded Technology and Applications Symposium, RTAS
2021-May
ISSN(印刷版)1545-3421

会议

会议27th IEEE Real-Time and Embedded Technology and Applications Symposium, RTAS 2021
Virtual, Online
时期18/05/2121/05/21

指纹

探究 'Brief industry paper: Optimizing memory efficiency of graph neural networks on edge computing platforms' 的科研主题。它们共同构成独一无二的指纹。

引用此