Summary
In this chapter, we learned about a theory that is helpful in simplifying and quantifying complex connected systems called networks. Graph theory is the study of relationships (represented as edges in graphs) between dynamic entities and helps better interpret network models. We further elaborated (with Python code) on how an optimization problem can be mathematically formulated and solved using this concept. A lot of problems can be approached using a graph framework that involves the components of mathematical optimization, as discussed in a section of this chapter.
This chapter also introduced GNNs, which operate on the structure and property of a graph. A single property is predicted for an entire graph for a graph-level task, a property of each node is predicted for a node-level task, and the property of each existing edge in a graph is predicted abstractly an edge-level task. GNNs are applied when graphs are complex and deep.
In the next chapter, we will study the...