[Insight-users] Regular Step Gradient Descent optimizer

[Martin Urschler]

hi

i only have a small comment on your parameters for the LBFGSB optimizer:
setting the bounds like the way you are doing it, IMHO has no effect.

 >    boundSelect.Fill(  0 );
 >    upperBound.Fill(  10.0 );
 >    lowerBound.Fill( -10.0 );

if I remember the docs of this optimizer correctly, the boundSelect 
variable is used to switch on the upper and/or lower bounds for the 
optimization per unknown, however if you select 0 for all entries in 
boundSelect, then all unknowns of your optimization will be UNBOUNDED -> 
as a consequence you may put whatever bounds you want into 
upper/lowerBound, it won't have any effect!

regards,
Martin


> The only change is the step length but the results are similar to the 
> previous one. By the way for the LBFGSB optimizer I used the following 
> conditions:
>  
>       // (1) LBFGSB optimizer
>    OptimizerType::BoundSelectionType boundSelect( 
> transform->GetNumberOfParameters() );
>    OptimizerType::BoundValueType upperBound( 
> transform->GetNumberOfParameters() );
>    OptimizerType::BoundValueType lowerBound( 
> transform->GetNumberOfParameters() );
>    boundSelect.Fill(  0 );
>    upperBound.Fill(  10.0 );
>    lowerBound.Fill( -10.0 );
>    optimizer->SetBoundSelection( boundSelect );
>    optimizer->SetUpperBound( upperBound );
>    optimizer->SetLowerBound( lowerBound );
>    optimizer->SetMaximumNumberOfEvaluations( 200 );
>    optimizer->SetMaximumNumberOfCorrections( 200 );
>  
> Here I am not clear what you meant for best setting. I thought the 
> conditions I provided were reasonable. Do you see anything obviously 
> different which leads to different results?


-- 
Martin Urschler
TU Graz - Institute for Computer Graphics & Vision
EMail: martin at urschler.info  ICQ: 41586273
Web: http://www.urschler.info
-----------------------------------------------------------
"Wehre nichts ab, was zu dir kommt, und halte nichts fest, was von dir 
geht."