[Insight-users] Gradient descent Question PLEASE I NEED HELP

[Yixun Liu]

Maybe you mean you do not know the gradient value added to the theta_t in
your last equation. The gradient part includes three terms: sum, partial
derivative with respect to T and the matrix. The sum can be calculated
because you know the theta_t. The derivative can be calculated by 1) finite
differencing or 2) interpolating a continuous Image I_m and derive its
analytical derivative. Both methods rely on the interpolator you use.



The first term relies on the metric. As I know SSD which you use, CC and MI
have analytical derivative. But there are lots of metrics without
analytical derivative. ITK uses finite differencing to calculate
derivative, so it can deal with any metrics although might lose some
performance for the SSD, CC and MI.


On Tue, Apr 30, 2013 at 9:51 AM, alaamegawer <alaamegawer at yahoo.com> wrote:

>
>
> i mean the constant value added to initialization theta0 ,because
> transformation matrix became constant i mean we multiply  difference *
> gradient * 1(from transformation matrix) then add this to theta0
>
>
>
> --
> View this message in context:
> http://itk-insight-users.2283740.n2.nabble.com/Gradient-descent-Question-PLEASE-I-NEED-HELP-tp7582905p7582916.html
> Sent from the ITK Insight Users mailing list archive at Nabble.com.
> _____________________________________
> Powered by www.kitware.com
>
> Visit other Kitware open-source projects at
> http://www.kitware.com/opensource/opensource.html
>
> Kitware offers ITK Training Courses, for more information visit:
> http://www.kitware.com/products/protraining.php
>
> Please keep messages on-topic and check the ITK FAQ at:
> http://www.itk.org/Wiki/ITK_FAQ
>
> Follow this link to subscribe/unsubscribe:
> http://www.itk.org/mailman/listinfo/insight-users
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.itk.org/pipermail/insight-users/attachments/20130430/a2d65151/attachment.htm>