Let us consider the following imaginary and simple example. Suppose that we want to automate the process of sorting cuts of white marble according to their origin: Carrara or Thassos. We will try to distinguish Carrara marble from Thassos marble using optical sensing.

Carrara        "Bianco"	Thassos    "Snow White"

A system to perform this very specific task might have the following form:

• A camera takes a picture of the stone, and passes the picture to a feature extractor.
  The purpose of this extractor is to measure certain "features" or "properties" which distinguish pictures of Carrara marble from pictures of Thassos marble.
  
  If we know by experience that Thassos marble is often lighter colored than Carrara marble, then brightness becomes an obvious feature, and we might attempt to classify the marble pieces merely by taking brightness into account.

• The values of these features are then passed to a classifier which evaluates the evidence presented and makes a final decision about the marble type.