arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2603.06952cs.LGcs.DB
收藏

并非所有邻居都重要:理解图稀疏化对GNN流程的影响

Not All Neighbors Matter: Understanding the Impact of Graph Sparsification on GNN Pipelines

Boston University(波士顿大学);Chalmers University of Technology(查尔姆斯理工大学)

Yuhang Song, Naima Abrar Shami, Romaric Duvignau, Vasiliki Kalavri

AI总结:

本文研究图稀疏化对GNN流程的影响,发现稀疏化能提升模型准确性并显著加快训练和推理速度。

摘要:

随着图规模扩展到数十亿个节点和边,图机器学习工作负载受到多跳遍历指数增长邻居的代价限制。尽管已经提出了各种系统级和算法级优化来加速图神经网络(GNN)流程,但数据管理和移动仍然是大规模下的主要瓶颈。在本文中,我们探讨图稀疏化,一种已建立的技术,通过减少边来创建更稀疏的邻居,是否可以作为轻量级的预处理步骤来解决这些瓶颈,同时在节点分类任务上保持准确性。我们开发了一个可扩展的实验框架,使能够系统评估不同稀疏化方法对GNN模型性能和准确性的影響。我们进行了首次全面研究GNN训练和推理在稀疏化图上的表现,揭示了几个关键发现。首先,稀疏化通常保持或甚至提高预测性能。例如,随机稀疏化在PubMed图上将GAT模型的准确性提高了6.8%。其次,收益随着规模增加而显著加快训练和推理。我们的结果表明,K-Neighbor稀疏化器在Products图上仅以0.7%的准确性下降,将模型服务性能提高了11.7倍。重要的是,我们发现稀疏化的计算开销很快被抵消,使其在非常大的图中变得实用。

英文摘要:

As graphs scale to billions of nodes and edges, graph Machine Learning workloads are constrained by the cost of multi-hop traversals over exponentially growing neighborhoods. While various system-level and algorithmic optimizations have been proposed to accelerate Graph Neural Network (GNN) pipelines, data management and movement remain the primary bottlenecks at scale. In this paper, we explore whether graph sparsification, a well-established technique that reduces edges to create sparser neighborhoods, can serve as a lightweight pre-processing step to address these bottlenecks while preserving model quality. We develop an extensible experimental framework that integrates graph sparsification into GNN training and inference pipelines, and use it to study how four representative sparsification methods interact with four GNN architectures, across six graphs and two learning tasks (node classification and link prediction). Our central finding is counter-intuitive: removing a large fraction of a graph's edges, often the majority, has only a small effect on model quality, and sometimes improves it. For example, the K-Neighbor sparsifier removes 83.3% of the edges of the Products graph while GCN accuracy drops by only 0.2 percentage points, and on the Arxiv link prediction task, sparsification raises GraphSAGE test AUC from 0.86 to 0.95. Because the runtime costs scale with the number of edges, these reductions translate into substantial speedups. Our results show that the pre-processing overhead is quickly amortized, making sparsification practical even for very large graphs.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑