Pattern verification is mathematically defined, an appropriate decision function derived, and a measure for system evaluation is given. Two basic postulates are set forth to fully define a verification system: each known class is expected, with nonzero probability, to be verified under the correct class label; and the pattern vector extracted during verification should be descriptive of the given class, independent of which class label was entered into the system. Through appropriate use of a priori probabilities, three types of information can be incorporated into the theory: the expected number of times a given class will require verification, the expected use of each class label by a given class, and the likelihood that a particular class is susceptible to “impostor” patterns.
Note: The Institute of Electrical and Electronics Engineers, Incorporated is distributing this Article with permission of the International Business Machines Corporation (IBM) who is the exclusive owner. The recipient of this Article may not assign, sublicense, lease, rent or otherwise transfer, reproduce, prepare derivative works, publicly display or perform, or distribute the Article.