"imran" <email@example.com> wrote in message
> DEAR ALL
> I AM WORKING TO ESTIMATE CLEANER TIME FREQUENCY DISTRIBUTIONS USING
> NEURAL NETWORKS
> I WANT TO INCORPORATE CRAMER RAO BOUND AS THE DECISION CRITERIA
> CAN SOMEONE PROVIDE ME WITH A GOOD UNDERSTANDING ON CRB MAY BE BY SOME
> GOOD ARTICLE OR APPLICATION?
> I SHALL BE OBLIGED
An estimate is a random variable that is a (typically nonlinear) function of
a set of other random variables. In many instances--I'm not sure if it is
all instances--if you linearize this function around a particular value you
can calculate the covariance of the estimate as a function of the covariance
of the set of random variables. This is the CR(L)B. More complete coverage
can be found in any book on estimation theory or by Googling "Cramer Rao".