LEARNING IN PATTERN RECOGNITION
Abstract
This technical note deals with the study of optimum learning in pattern recognition problems. The class of pattern recognition problems which is treated is perfectly general and the results may be appled to patterns of visual, aural or electromagnetic origin. Mathematical solutions are derived and practical instrumentation of these solutions are suggested for the following pattern recognition situations: (1) recognition of fixed a priori known patterns immersed in additive gaussian noise, (2) recognition of unknown fixed patterns immered in additive gaussian noise, (3) recognition of patterns drawn from a gaussian ensemble with known mean and covariance and immersed in additive gaussian noise, (4) learning in pattern recognition, (5) learning without a teacher, and (6) recognition of unknown patterns which vary in time (pattern tracking). (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Oct 01, 1961
- Accession Number
- AD0267798
Entities
People
- N. Abramson
Organizations
- ITT Corporation