BriefGPT.xyz
May, 2019
再探图神经网络：一切都是低通滤波器
Revisiting Graph Neural Networks: All We Have is Low-Pass Filters
HTML
PDF
Hoang NT, Takanori Maehara
TL;DR
本文研究了基于图形信号处理的理论框架，用于分析图神经网络的性能，结果表明，图神经网络仅对特性向量进行低通滤波，且没有非线性流形学习属性。进一步研究了它们对特征噪声的鲁棒性，并提出了一些基于GCN的图神经网络设计洞见。
Abstract
graph neural networks
have become one of the most important techniques to solve machine learning problems on graph-structured data. Recent work on
vertex classification
proposed deep and distributed learning mode
→