GNN_learning

This is a repo about GNN and DGL

GNN 入门

代码

• 安装wsl2（windows的linux子系统）
• 配python环境+安装DGL Deep Graph Library (dgl.ai)
• 读DGL里面三个模型GCN, RGCN, GAT的代码，跑一下example

论文

• 下载zotero

• A Gentle Introduction to Graph Neural Networks (distill.pub)

• 综述论文：Graph Neural Networks: A Review of Methods and Applications

• 了解GNN的基本模型：GCN (Graph convolution network), RGCN (Relational graph convolution network), GAT (Graph attention network)

文献调研

总结整合了GNN目前的主流研究方向

汇总目前GNN比较热门的研究方向

• Graph Generation & Transformation
• Dynamic Graph
• Graph Matching
• GNN sample methods
• Hetergeneous graph
• Link prediction

reference:

recently read paper

UNDERSTANDING GNN COMPUTATIONAL GRAPH

model

GCN (Graph convolutional network)

• Semi-Supervised Classification with Graph Convolutional Networks
• Graph Convolutional Network — DGL 0.8.0post2 documentation

GAT (Graph attention network)

• Graph Convolutional Network (GCN)
• Understand Graph Attention Network — DGL 0.8.0post2 documentation

RGCN (Relational graph convolutional network)

• Modeling Relational Data with Graph Convolutional Networks
• Relational Graph Convolutional Network — DGL 0.8.0post2 documentation

TGN (temporal graph network)

• Accelerating and scaling Temporal Graph Networks on the Graphcore IPU

• twitter-research/tgn: TGN: Temporal Graph Networks (github.com)

DGL

build-in function in DGL

Notebooks:

reduce and message function in DGL