《A Gentle Introduction to Graph Neural Networks》

Reference：

A Gentle Introduction to Graph Neural Networks

Understanding Convolutions on Graphs


 

A graph represents the relations (edges) between a collection of entities (nodes).

 we can store information in each of these pieces of the graph.

Images as graphs

 Text as graphs

Graph-level task

predict the property of an entire graph

 Node-level task

Node-level tasks are concerned with predicting the identity or role of each node within a graph.

Edge-level task

The remaining prediction problem in graphs is edge prediction.

nodes, edges, global-context and connectivity

nodes, edges, global-context：向量表示

connectivity：连接矩阵（非常大）/稀疏矩阵（不能放GPU）表示都有弊端

Graph Neural Networks

graph-in, graph-out

The simplest GNN

 GNN Predictions by Pooling Information

 Pooling proceeds in two steps:

1. For each item to be pooled, gather each of their embeddings and concatenate them into a matrix.
2. The gathered embeddings are then aggregated, usually via a sum operation.

 If we only have edge-level features, and are trying to predict binary node information：

 If we only have node-level features，and are trying to predict binary edge-level information：

 If we only have node-level features, and need to predict a binary global property：

 An end-to-end prediction task with a GNN model：

 Passing messages between parts of the graph

Other types of graphs：

Sampling Graphs and Batching in GNNs