Accession Number:

AD0416795

Title:

PATTERN SEPARATION BY CONVEX PROGRAMMING,

Descriptive Note:

Corporate Author:

STANFORD UNIV CALIF APPLIED MATHEMATICS AND STATISTICS LABS

Personal Author(s):

Report Date:

1963-06-28

Pagination or Media Count:

20.0

Abstract:

It is shown that the pattern separation problem can be formulated and solved as a convex pro gramming problem, i.e., the minimization of a convex function subject to linear constraints. A number of previous investigators have proposed iterative methods for the construction of one or more hyperplanes in order s solve pattern recognition problems. It was apparently not recognized that these iterative methods were, in fact, ining a feasible solution to a mathematical programming problem. Very effi cient computer methods have been developed for such programming problems and can be used to advantage for the pattern recognition problem. Author

Subject Categories:

Distribution Statement:

APPROVED FOR PUBLIC RELEASE