A classifier, once estimated from the features found in samples of data, assigns probabilities to classes on future samples when evaluated.

A generative classifier builds models of the form P(class, feature).