Supervised learning methods
This section describes the concrete supervised learning methods that can be applied in cybersecurity scenarios. We’ll describe where they can be applied, how they work, and what their advantages and disadvantages are. Supervised learning methods are used to solve problems where we have input data (X) and target outputs (Y), which means the training and validation data needs to be labeled with its associated output. In cybersecurity, an example would be data that is labeled as an attack or benign network traffic. The goal is to train the machine learning model so that it can find the pattern that produces the target output.
In other words, we are looking for a mathematical function (f) with a parameter set (W), where Y=f(X, W). For instance, a simple example would be a linear function, Y=WX. Depending on our assumptions, we can use different types of functions, such as linear, polynomial, sine, and so on. During training, the parameters of the function...
