[GNN] 2. GNN 이전의 Machine Learning을 활용한 Graph 학습

 

[GNN] 2. GNN 이전의 Machine Learning을 활용한 Graph 학습

 

 

 

1. Graph란?

- 왜 Graph?

- Graph의 종류

- Graph의 표현

- Graph Tasks

- Graph의 Motif 

 

 

2. GNN 이전의 Machine Learning을 활용한 Graph 학습 

- Node Feature

• Eigenvector centrality
• Betweenness centrality
• Closeness centrality
• .....

- Link Feature

• Distance-based feature
• Local neighborhood overlap
• Global neighborhood overlap

- Graph Feature

• Graphlet kernel
• Weisfeiler-Lehman Kernel 

 

3. Graph Representation Learning

- Node Embedding

- Graph Embedding 

 

 

4. GNN

- Recurrent Graph Neural Networks (RecGNNs)

- Convolutional Graph Neural Networks (ConvGNNs)

• Spectral Model
• Spatial Model 

- Graph AutoEncoders (GAEs) 

• Network Embedding
• Graph Generation

- Spatial-Temporal Graph Neural Networks (STGNNs)

• CNN Based 
• RNN Based
• CNN & RNN based Model 
• Attention based Model 

 

5. GNN의 Benchmark Datasets

- Citation Networks

- Biochemical Graphs

- Social Networks 

- Others

 

 

6. GNN Library

- Pytorch Geometric

- DGL 

 

 

 

---

 

 

Node-level Feature

 

 

 

Link-level Feature

 

 

 

 

Graph-level Feature

 

 

 

 

 

 

 

Reference

 

https://velog.io/@tobigsgnn1415/Traditional-Methods-for-Machine-Learning-in-Graphs

2. Traditional Methods for Machine Learning in Graphs

https://harryjo97.github.io/theory/Weisfeiler-Lehman-Algorithm/

Weisfeiler-Lehman Algorithm

https://davidbieber.com/post/2019-05-10-weisfeiler-lehman-isomorphism-test/