If we know Bayesian statistics , we can understand the difference of generative model and discriminative model in pattern recognition .
Discriminative model is the method to estimate "P (Y|X)" directly.
Logistic Regression Analysis is an example. We can get the value of probability of Y from X.
Generative model is the method to estimate "P (Y|X)" from "P (X|Y)", P (Y) and P (X).
Discriminative model is useful for pattern recognition because it is simpler.
Generative model is used if P (X|Y) is wanted to use.
It is used to make data or to increase data for Artificial Intelligence.