Generative, Descriptive, Heuristic, and Formal Modeling in Pattern Analysis and Classification.

Abstract

Generative grammars and pattern descriptions have been recently associated primarily with the Linguistic School of pattern recognition research. The author shows that very similar types of generative and descriptive models are used extensively in obtaining stochastic models for patterns. The pure linguistic model is usually presented as a contrast to the pure statistical classification model. Neither pure model is relevant to applications. Rather it is hybrid linguistic-statistical approaches that have been useful in practice. The author examines the relationship of measurement analysis to pattern classification, formalisms for pattern analysis and grammatical inference, heurristic transformations for inferring pattern grammars, and the role which heuristics play in making a formalism successful. The author presents an example of the generative and descriptive modeling of error-clusters and error-gaps which occur in the transmission of binary data over digital communication channels with memory and conclude with a discussion of the testing of theories and models. (Author)

Document Details

Document Type
Technical Report
Publication Date
Jul 01, 1971
Accession Number
AD0730285

Entities

People

  • Laveen N. Kanal

Tags

DTIC Thesaurus Topics

  • Classification
  • Communication Channels
  • Computer Science
  • Computers
  • Contrast
  • Cooperation
  • Digital Communications
  • Maryland
  • Measurement
  • Pattern Recognition
  • Pennsylvania
  • Recognition
  • Theoretical Computer Science

Readers

  • Computational Linguistics
  • Neural Network Machine Learning.

Technology Areas

  • AI & ML
  • AI & ML - Machine Translation