A PROBLEM IN PATTERN RECOGNITION,

Abstract

Independent observations, made simultaneously on members of each of two categories, are presented to an artificial intelligence. It is assumed that the space of the attribute or observation vectors of the individuals to be classified is a finite dimensional Euclidean space, and that the attribute vectors are strictly separable by a hyperplane. When the artificial intelligence is presented with a pair of vectors, it is informed of the categories to which these vectors belong; it then estimates a separating hyperplane. This paper describes an algorithm for making this estimate and gives conditions for the convergence of this estimate to a hyperplane which correctly separates the regions. In general, this type of learning problem may be solved by any of a large class of algorithms which differ in their convergence rates, complexity of computation and amount of memory. It is hoped that the relatively simple convergence proof for the algorithm of this paper will provide some insight into the more general problem.

Document Details

Document Type
Technical Report
Publication Date
Jun 01, 1964
Accession Number
AD0604370

Entities

People

  • Zivia S. Wurtele

Organizations

  • University of California, Los Angeles

Tags

DTIC Thesaurus Topics

  • Algorithms
  • Artificial Intelligence
  • Computational Complexity
  • Computations
  • Convergence
  • Identification
  • Learning
  • Mathematical Analysis
  • Mathematics
  • Observation
  • Pattern Recognition
  • Recognition

Fields of Study

  • Mathematics

Readers

  • Geospatial Intelligence and Artificial Intelligence Analytics
  • Graph Algorithms and Convex Optimization.
  • Theoretical Analysis.

Technology Areas

  • AI & ML
  • AI & ML - Machine Learning Algorithms
  • Space