So far we’ve discussed methods in some generality without touching much on how we might use the methods in a practical setting. Here we discuss one of the simplest methods that is widely used in practice to classify data. This is a useful junction since it enables us to discuss the issues of parameter learning from data and also (constrained) structure learning.
10.1 Naive Bayes and conditional independence
We shall discuss machine learning concepts in some detail in Chapter 13. Here, we require only the intuitive concept of classification, which means giving a discrete label to an input. For example, one might wish to classify an input image into one of two classes – male or female. Naive Bayes (NB) is a popular classification ...