Learning and Evaluation of Pattern Characteristics for Automatic Recognition.
Abstract
The design of an automatic pattern recognition system involves considerations in three distinct but not totally independent problem areas. The broad selection of pattern characteristics and the design of equipment which will measure these characteristics quickly and efficiently are two aspects of the system which must be resolved. The third is the formulation of a recognition scheme which will establish the significant characteristics and operate upon them to form reliable pattern classifications. This report is concerned primarily with the classification problem and discusses some applications of statistical decision theory and information theory to the design of a recognition scheme. One section of the report presents the results of an experiment and the other problem areas which were considered -- to the extent that characteristics had to be defined and measured before the data could be collected.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jan 01, 1962
- Accession Number
- ADA053841
Entities
People
- G. G. Ricker