DocumentCode :
921766
Title :
Finite-memory algorithms for estimating the mean of a Gaussian distribution (Corresp.)
Author :
Hellman, M.E.
Volume :
20
Issue :
3
fYear :
1974
fDate :
5/1/1974 12:00:00 AM
Firstpage :
382
Lastpage :
384
Abstract :
Let 
be independent random variables, each having a 
distribution. If we try to estimate 
with an 
-state learning algorithm, then the minimum mean-squared error is bounded below by that obtained by the best -level quantizer (which requires knowledge of ). Here we show that this lower bound is tight. The results are easily extended to a number of other problems, such as estimating the mean 
of a uniform distribution.
be independent random variables, each having a distribution. If we try to estimate with an -state learning algorithm, then the minimum mean-squared error is bounded below by that obtained by the best -level quantizer (which requires knowledge of ). Here we show that this lower bound is tight. The results are easily extended to a number of other problems, such as estimating the mean of a uniform distribution.Keywords :
Finite-memory methods; Gaussian processes; Parameter estimation; Acoustic signal detection; Calibration; Error analysis; Estimation error; Gaussian distribution; Gaussian noise; Radar detection; Random variables; Signal to noise ratio; Statistical distributions;
fLanguage :
English
Journal_Title :
Information Theory, IEEE Transactions on
Publisher :
ieee
ISSN :
0018-9448
Type :
jour
DOI :
10.1109/TIT.1974.1055229
Filename :
1055229
Link To Document :
