MULTIVARIATE DATA-ANALYSIS BASED ON THE W(N)(K)-CRITORIA AND MULTILAYER PERCEPTRON

Authors
BONUSHKINA AY IVANOV VV ZRELOV PV
Citation
Ay. Bonushkina et al., MULTIVARIATE DATA-ANALYSIS BASED ON THE W(N)(K)-CRITORIA AND MULTILAYER PERCEPTRON, Computers & mathematics with applications, 34(7-8), 1997, pp. 677-685
Citations number
10
Categorie Soggetti
Computer Sciences",Mathematics,"Computer Science Interdisciplinary Applications
Journal title
Computers & mathematics with applications → ACNP
ISSN journal
08981221
Volume
34
Issue
7-8
Year of publication
1997
Pages
677 - 685
Database
ISI
SICI code
0898-1221(1997)34:7-8<677:MDBOTW>2.0.ZU;2-H
Abstract
A comparative study of multidimensional classifiers based on the goodn ess-of-fit criteria omega(n)(k) and multilayer perceptrons (MLP) has b een carried out. It is shown that MLP exhibits the ''instantaneous'' l earning effect and improves the quality of recognition in the case of input data represented in the form of variational series. The reasons are analyzed that underlie these effects. Recommendations for joint us age of the omega(n)(k) criteria and of MLPs are given.