Accession Number : AD0704651

Title :   PATTERN CLASSIFICATION WITH A PARTITIONED TRAINING SET,

Corporate Author : ILLINOIS UNIV URBANA COORDINATED SCIENCE LAB

Personal Author(s) : Chow,James Cheh-Min

Report Date : APR 1970

Pagination or Media Count : 36

Abstract : Pattern classification can be considered as consisting of two parts: (1) Pattern detection - The process of learning from a set of sample patterns of known classifications and discriminating characteristics of each category; and (2) Actual classification - The process of recognizing patterns of unknown classifications as members of particular categories. The paper is a study in the first part of the process since it is most often te more important part of any pattern classification scheme. An algorithm for establishing decision criteria of classification is described. Evaluation is made on its performance, computation time and data storage requirement. (Author)

Descriptors :   (*PATTERN RECOGNITION, CLASSIFICATION), DECISION THEORY, LEARNING, ITERATIONS, ALGORITHMS

Subject Categories : Bionics

Distribution Statement : APPROVED FOR PUBLIC RELEASE