Accession Number : AD0705674

Title :   APPLICATION OF WIENER CANONICAL EXPANSIONS TO PATTERN RECOGNITION. VOLUME I.

Descriptive Note : Final rept. 18 Oct 65-31 Dec 69,

Corporate Author : INFOTON INC BURLINGTON MASS

Personal Author(s) : Brick,Donald B. ; Henrichon,Ernest G. ; Citron,Alan L. ; Stout,R. W.

Report Date : 30 JAN 1970

Pagination or Media Count : 39

Abstract : The investigations centered upon the following areas of research: An examination of the existence, convergence, and ergodicity properties of Wiener Canonical Expansions of Bayes rules, for the purpose of specifying the class of stochastic processes that can be treated by this procedure; A study of the effects of truncating the expansions to enable recognition of stochastic processes in a practical, physically realizable form; A comparison of the method with other approaches for describing the recognition of temporal waveforms; Extensions of the basic expansion procedure for synthesizing nonlinear systems to multiple input-output systems and discrete-logic systems. (Author)

Descriptors :   (*PATTERN RECOGNITION, STOCHASTIC PROCESSES), (*INFORMATION THEORY, *SPEECH RECOGNITION), ARTIFICIAL INTELLIGENCE, LEARNING MACHINES, ADAPTIVE SYSTEMS, MEASURE THEORY, NONLINEAR SYSTEMS

Subject Categories : Cybernetics
      Bionics
      Voice Communications

Distribution Statement : APPROVED FOR PUBLIC RELEASE