Getting started with Graph Convolutional Networks as an Amateur
Getting started with Graph Convolutional Networks as an Amateur https://zhuanlan.zhihu.com/p/27587371 https://www.cnblogs.com/wangxiaocvpr/p/8306519.html https://www.cnblogs.com/wangxiaocvpr/p/8299336.html https://towardsdatascience.com/how-to-do-deep-learning-on-graphs-with-graph-convolutional-networks-7d2250723780 GCN是一种强大的神经网络,旨在直接在图上工作并利用其结构信息。这篇文章是关于如何用图卷积网络(GCN)在图上进行深度学习的系列文章的第一篇。该系列的包含: 本篇:一个高层级的GCN介绍 谱图卷积的半监督学习 最基本的代码实现GCN – 一个高层级的GCN介绍 In this post, I will illustrate how information is propagated through the hidden layers of a GCN using coding examples. 看看GCN是如何从前几层聚合信息的,以及这种机制是如何产生图中节点的有用特征表示的。 ref link in Chinese中文 看看GCN的输入是啥 Given a graph $G = (V, E)$ , a GCN takes as input: adjacency matrix $\mathbf{A}$ of $G$. N × N an input feature matrix, $N × F⁰$ feature matrix, $\mathbf{X}$, where N is the number of nodes and F⁰ is the number of input features for each node 图示是一个例子
…