Robust methodology for improving the conditionedness of correlation matrices

Authors
Citation
Ta. Aliev, Robust methodology for improving the conditionedness of correlation matrices, AUTOM C C S, 35(1), 2001, pp. 8-20
Citations number
8
Categorie Soggetti
AI Robotics and Automatic Control
Journal title
AUTOMATIC CONTROL AND COMPUTER SCIENCES
ISSN journal
01464116 → ACNP
Volume
35
Issue
1
Year of publication
2001
Pages
8 - 20
Database
ISI
SICI code
0146-4116(2001)35:1<8:RMFITC>2.0.ZU;2-M
Abstract
A robust methodology for improving the conditionedness of correlation matri ces is proposed. The method makes it possible to eliminate the effect of os cillations in the noise characteristics on the error of the matrix elements and to balance the relative errors in the matrix elements. Through the use of robust estimators thus obtained it becomes possible to transform the in itial matrix so that it assumes the form of a matrix whose elements do not contain noise errors by means of the technique of balancing the relative er rors.