You're reading from Hands-On Graph Neural Networks Using Python

Product type Book

Published in Apr 2023

Publisher Packt

ISBN-13 9781804617526

Pages 354 pages

Edition 1st Edition

Languages

Concepts

Neural Networks

Author (1):

Maxime Labonne

Table of Contents (25) Chapters

Preface

1. Part 1: Introduction to Graph Learning

2. Chapter 1: Getting Started with Graph Learning

3. Chapter 2: Graph Theory for Graph Neural Networks

4. Chapter 3: Creating Node Representations with DeepWalk

5. Part 2: Fundamentals

6. Chapter 4: Improving Embeddings with Biased Random Walks in Node2Vec

7. Chapter 5: Including Node Features with Vanilla Neural Networks

8. Chapter 6: Introducing Graph Convolutional Networks

9. Chapter 7: Graph Attention Networks

10. Part 3: Advanced Techniques

11. Chapter 8: Scaling Up Graph Neural Networks with GraphSAGE

12. Chapter 9: Defining Expressiveness for Graph Classification

13. Chapter 10: Predicting Links with Graph Neural Networks

14. Chapter 11: Generating Graphs Using Graph Neural Networks

15. Chapter 12: Learning from Heterogeneous Graphs

16. Chapter 13: Temporal Graph Neural Networks

17. Chapter 14: Explaining Graph Neural Networks

18. Part 4: Applications

19. Chapter 15: Forecasting Traffic Using A3T-GCN

20. Chapter 16: Detecting Anomalies Using Heterogeneous GNNs

21. Chapter 17: Building a Recommender System Using LightGCN

22. Chapter 18: Unlocking the Potential of Graph Neural Networks for Real-World Applications

23. Index

Why subscribe?

24. Other Books You May Enjoy

Comparing graph convolutional and graph linear layers

In the previous chapter, our vanilla GNN outperformed the Node2Vec model, but how does it compare to a GCN? In this section, we will compare their performance on the Cora and Facebook Page-Page datasets.

Compared to the vanilla GNN, the main feature of the GCN is that it considers node degrees to weigh its features. Before the real implementation, let’s analyze the node degrees in both datasets. This information is relevant since it is directly linked to the performance of the GCN.

From what we know about this architecture, we expect it to perform better when node degrees vary greatly. If every node has the same number of neighbors, these architectures are equivalent: ():

We import the Planetoid class from PyTorch Geometric. To visualize the node degrees, we also import matplotlib and two additional classes: degree to get the number of neighbors of each node and Counter to count the number of nodes for each...