INDUCTIVE PROCESSES FOR FACTOR SELECTION IN A RECOGNITION MODEL

Abstract

Given a set of measured characteristics for an unknow object, it i esired to assign this object to a class, the implicatio s of hich re known. Intuitively, the set of measured characteristics sufficient to achieve this assignment must be a specialized set, dependent on the nature and umber of categories, and on the natre of the objects being considered. The theoretical model and its application presented show that t e requireme ts for a set of parameters to be used as inputs are not as rigid as originally assumed. It is sho tha a lessqualified set of measured inputs can be u o i uce an expanded set which combines bot natural (m asured) and artificial parameters. The expanded set allows the effective definition of a recognition function regardless of whether or not the original measur ments encoded r sufficient in themselves to provide the completed recognition model. (Author)

Document Details

Document Type
Technical Report
Publication Date
Dec 01, 1961
Accession Number
AD0270013

Entities

People

  • Richard M. Blood

Tags

DTIC Thesaurus Topics

  • Recognition

Readers

  • Computational Modeling and Simulation
  • Neural Network Machine Learning.
  • Snow Cover Descriptors for Reptiles and Their Illustrations.