Classification is the task of assigning a particular input to the class that it belongs to and is an important component of many problem solving tasks.
The task of constructing a class definition is called concept learning or induction. A number of algorithms exist which automatically generate class definitions. These algorithms usually require a set of pre-classified, negative and positive, training examples of the concept or class. Once generated the class definition may be used to determine if a as yet unclassified example belong to the class or not.
Algorithms that require a set of pre-classified examples are referred to as supervised learning algorithms as opposed to un-supervised when this is not a requirement. A further distinction can be made between algorithms that require all training examples to be present before learning (these are referred to as non-incremental learning algorithms) and those that can process and refine their class definitions as new examples become available. These algorithms are referred to as incremental algorithms.
Authored by Serengul Smith
E-mail to:
serengul1@mdx.ac.uk
School of Computing Science Middlesex University
Revised: September 1998