Summary

In this chapter, we answered three main questions: why graphs, why graph learning, and why graph neural networks? First, we explored the versatility of graphs in representing various data types, such as social networks and transportation networks, but also text and images. We discussed the different applications of graph learning, including node classification and graph classification, and highlighted the four main families of graph learning techniques. Finally, we emphasized the significance of GNNs and their superiority over other techniques, especially regarding large, complex datasets. By answering these three main questions, we aimed to provide a comprehensive overview of the importance of GNNs and why they are becoming vital tools in machine learning.

In Chapter 2, Graph Theory for Graph Neural Networks, we will dive deeper into the basics of graph theory, which provides the foundation for understanding GNNs. This chapter will cover the fundamental concepts of graph theory, including concepts such as adjacency matrices and degrees. Additionally, we will delve into the different types of graphs and their applications, such as directed and undirected graphs, and weighted and unweighted graphs.