Summary
In this chapter, we explored SVM as a classifier. In addition to linear data, SVMs can efficiently classify non-linear data using kernel functions. The method used by the SVM algorithm can be extended to solve regression problems. SVM is utilized for novelty detection as well, wherein the training dataset is not polluted with outliers and the algorithm is exploited to detect a new observation as an anomaly, in which case the outlier is called a novelty.
The next chapter is about graph theory, a tool that provides the necessary mathematics to quantify and simplify complex systems. Graph theory is the study of relations (connections or edges) between a set of nodes or individual entities in a dynamic system. It is an integral component of ML and DL because graphs provide a means to represent a business problem as a mathematical programming task in the form of nodes and edges.