[Insight-developers] Questions about ParamterScaleEstimator

[Bill Lorensen]

Why can't this also be estimated? When I run the test, I tried several
different values without success. To me, it is a deal breaker if I have to
play with this free parameter.

Bill

On Fri, May 6, 2011 at 12:41 PM, Baohua Wu <baowu99 at gmail.com> wrote:

>  Bill,
>
> This factor is like the learningRate in GradientDescentOptimizer. It is
> roughly chosen based on the gradient in the optimizer such that a step won't
> cause too much voxel shift. Thank you!
>
> Bao
>
> ----- Original Message -----
> *From:* Bill Lorensen <bill.lorensen at gmail.com>
> *To:* baowu99 at gmail.com
> *Cc:* Insight Developers <insight-developers at itk.org>
> *Sent:* Friday, May 06, 2011 8:23 AM
> *Subject:* Questions about ParamterScaleEstimator
>
> Bao,
>
> I am reviewing you ParameterScaleEstimator class in gerrit:
> http://review.source.kitware.com/#change,1499
>
> I notice that the test sets a global scaling factor of 3.5e+7. What is the
> purpose of this number and how did you pick that value. If I modify the test
> and do not set it, the test fails.
>
> Bill
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.itk.org/mailman/private/insight-developers/attachments/20110506/d00bc6e1/attachment.htm>