RECOGNITION BY CONSTRUCTIVE NEURAL ALGORITHMS

Authors
DIZENZO S BURGESS N FERRAGINA P GRANIERI MN
Citation
S. Dizenzo et al., RECOGNITION BY CONSTRUCTIVE NEURAL ALGORITHMS, Pattern recognition letters, 14(12), 1993, pp. 997-1007
Citations number
28
Categorie Soggetti
Computer Sciences, Special Topics","Computer Science Artificial Intelligence
Journal title
Pattern recognition letters → ACNP
ISSN journal
01678655
Volume
14
Issue
12
Year of publication
1993
Pages
997 - 1007
Database
ISI
SICI code
0167-8655(1993)14:12<997:RBCNA>2.0.ZU;2-G
Abstract
The usability of the constructive neural algorithms as pattern classif iers is investigated. It is pointed out that the unboundedness of the decision regions formed by most neural recognizers leads to substantia l limitations of the generalization capabilities of these nets. We spe cify a constructive neural recognizer that forms bounded decision regi ons, and report the results of this algorithm on a series of benchmark problems that resemble the usual pattern recognition problems.