Title :
A criterion for selecting information- preserving data reductions for use in the design of multiple- parameter estimators (Corresp.)
Author :
Bruzzone, Stephen P. ; Kaveh, Mostafa
fDate :
5/1/1983 12:00:00 AM
Abstract :
A scalar measure of the fractional amount of the total information for parameter estimation that is retained when a record of raw data is reduced to a given statistic is introduced. This measure, the relative information index (RII), is a function of Fisher´s information matrices for the statistic and for the raw data, and it resembles an efficiency measure in that it acquires values between zero and one. A criterion, based on the RII, is formulated for statistic selection, and some applications are discussed.
Keywords :
Parameter estimation; Application software; Noise reduction; Notice of Violation; Parameter estimation; Signal design; Signal detection; Signal processing; Signal to noise ratio; Statistics; Working environment noise;
Journal_Title :
Information Theory, IEEE Transactions on
DOI :
10.1109/TIT.1983.1056657