Accession Number : AD0756495

Title :   On Discrete Variables in Pattern Recognition.

Descriptive Note : Final technical rept.,

Corporate Author : SYRACUSE UNIV N Y DEPT OF ELECTRICAL AND COMPUTER ENGINEERING

Personal Author(s) : Stoffel,James

Report Date : JAN 1973

Pagination or Media Count : 141

Abstract : DISCRIMINATE ANALYSIS, DISCRIMINANT FUNCTIONSThe central problem dealt with in the report concerns pattern recognition tasks which involve discrete measurement variables. In the first chapter, the particular difficulties associated with discrete variables in the pattern recognition discipline are illuminated. The next chapter presents historical evidence of the variety of techniques applied to the task of dealing with discrete variables in the context of a classification problem. The third chapter provides a derivation of the aposteriori probability distribution for the error rate in a discrete variable classification problem. In Chapter 4, the theoretical foundation for the proposed procedure of dealing with discrete variable classifier design tasks is presented. Chapter 5 contains an analysis of the proposed method of applying the theory of prime events and an example application of the method to a discrete variable design task. (Author Modified Abstract)

Descriptors :   (*PATTERN RECOGNITION, CLASSIFICATION), SAMPLING, PROBABILITY DENSITY FUNCTIONS, STOCHASTIC PROCESSES, MULTIVARIATE ANALYSIS, BINARY ARITHMETIC, INFORMATION THEORY, WEATHER FORECASTING, RANDOM VARIABLES, COMPUTER PROGRAMMING, ALGORITHMS

Subject Categories : Computer Programming and Software
      Bionics

Distribution Statement : APPROVED FOR PUBLIC RELEASE