• A probabilistic framework for solving classification problems. Bayes classifier represents an example of generative approach. As the name implies, it is based on Bayes Theorem.
Bayesian Classifier Scenario
• We assume there are d features:
• These features are measured on objects from two classes:
• According to Bayes’ theorem, the probability that the
observation vector belongs to
class is given by the following relationship:
Naïve Bayes (NB) Classifier
• A Bayes classifier is termed as the NB classifier when we assume that features are independent of each other. Such an assumption rarely holds true, but the assumption helps simplify math and the classifier design. The NB classifier is particularly popular for text classification.
• With independent features, the expression for posteriori computation
from the previous slide can be written as follows: