arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2502.00282cs.LG
收藏

GraphMinNet:具有轻量复杂度极简架构的图依赖学习

GraphMinNet: Learning Dependencies in Graphs with Light Complexity Minimal Architecture

University of Kentucky(肯塔基大学);University of California, Santa Barbara(加利福尼亚大学圣巴巴拉分校);Institute for Biomedical Informatics(生物医学信息学研究所)

Md Atik Ahamed, Andrew Cheng, Qiang Ye, Qiang Cheng

AI总结:

本文提出GraphMinNet,一种基于极简门控循环单元的图神经网络架构,通过线性计算复杂度高效建模长距离依赖,结合特征与位置编码提升表达能力,在十个数据集上取得领先性能。

摘要:

图神经网络(GNNs)在各种应用中展现出卓越的成功,但它们往往难以有效捕捉长距离依赖(LRD)。本文提出了GraphMinNet,一种新颖的GNN架构,将极简门控循环单元(Gated Recurrent Units)的思想推广至图结构数据。该方法实现了高效的LRD建模,具有线性计算复杂度,同时保持排列等变性和稳定性。该模型通过特征编码与位置编码的独特结合,融合了结构与位置信息,其表达能力被证明强于1-WL测试。理论分析表明,GraphMinNet在长距离上保持非衰减梯度,确保了有效的长距离信息传播。在包括分子图、图像图和合成网络在内的十个多样化数据集上的大量实验表明,GraphMinNet在计算高效的同时达到了最先进的性能。结果显示,其在10个数据集中的6个上表现优异,在其他数据集上取得有竞争力的结果,验证了该方法在捕捉局部和全局图结构方面的有效性。

英文摘要:

Graph Neural Networks (GNNs) have demonstrated remarkable success in various applications, yet they often struggle to capture long-range dependencies (LRD) effectively. This paper introduces GraphMinNet, a novel GNN architecture that generalizes the idea of minimal Gated Recurrent Units to graph-structured data. Our approach achieves efficient LRD modeling with linear computational complexity while maintaining permutation equivariance and stability. The model incorporates both structural and positional information through a unique combination of feature and positional encodings, leading to provably stronger expressiveness than the 1-WL test. Theoretical analysis establishes that GraphMinNet maintains non-decaying gradients over long distances, ensuring effective long-range information propagation. Extensive experiments on ten diverse datasets, including molecular graphs, image graphs, and synthetic networks, demonstrate that GraphMinNet achieves state-of-the-art performance while being computationally efficient. Our results show superior performance on 6 out of 10 datasets and competitive results on the others, validating the effectiveness of our approach in capturing both local and global graph structures.

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。

↑