Re: Cost Functions

On 22 Mai, 11:54, "mikedoesnotknow" <michael.sou...@xxxxxx> wrote:
Hi all,

I would like to fit a n-dimensional linear model to data. I know how to
obtain the weights of my estimator according to the minimum least squares
solution for a large number of samples.
Now, I was wondering whether there any cost functions that shape the error
according to a specified frequency behaviour, e.g. the error at low
frequencies is supposed to be small whereas the error at high frequencies
is permitted to be large.
I would be very grateful for any kind of reference or keywords.

Seems as a weighted least squares.

Rune
.

Relevant Pages

  • Re: Cost Functions
    ... obtain the weights of my estimator according to the minimum least squares ... frequencies is supposed to be small whereas the error at high frequencies ...
    (comp.dsp)
  • Re: Cost Functions
    ... obtain the weights of my estimator according to the minimum least squares ... frequencies is supposed to be small whereas the error at high frequencies ...
    (comp.dsp)
  • Cost Functions
    ... obtain the weights of my estimator according to the minimum least squares ... frequencies is supposed to be small whereas the error at high frequencies ...
    (comp.dsp)
  • Proposed CPAN module Statistics::LineFit
    ... statistics, the predicted y values and the residuals of the y ... weights are input in a separate array. ... method to check the status of the regression. ... regress- Do the least squares line fit ...
    (comp.lang.perl.modules)
  • Re: Polynomial Fit Program
    ... least squares fit program? ... In the simplest case (polynomials of moderate degree), ... If you have weights, e.g. w_i, then first formally ... RWTH - Aachen University ...
    (sci.math.num-analysis)