[BioC] Neural networks.

[Steve Lianoglou]

Hi,

On Sep 3, 2009, at 4:40 AM, Marcus Gry wrote:

> Dear list members.
>
> I am trying to make a sensitivity analysis (as derived by Zurada  
> 1994, Engelbrecht 1995) of input parameters (gene expression data)  
> when applying a neural network to classify different cancer  
> subtypes. Since I am no expert in the field, (rather a newbie), I  
> wonder if there exists an implementation in R that can be used to  
> measure the relative importance of the input variables for the  
> neural network. I read that the sensitivity matrix is the Jacobian  
> matrix of the output parameters over the input parameters. I think  
> that somehow I can grasp the concept, I just don’t know how to  
> implement it in R. Any help, or guidelines would be greatly  
> appreciated.

First suggestion is to check out the machine learning view on cran:

http://cran.r-project.org/web/views/MachineLearning.html

You'll see where to get a neural network implementation there from  
there.

Are you can calculate a jacobian matrix? Try searching the R-help list  
in gmane/nabble:

http://www.nabble.com/forum/Search.jtp?forum=13819&local=y&query=jacobian

You might also get more help from the general R-help mailing list ...  
even though your specific application is related to biological data, I  
think your questions might be better addressed there since there  
somehow more general in nature.

-steve

--
Steve Lianoglou
Graduate Student: Computational Systems Biology
   |  Memorial Sloan-Kettering Cancer Center
   |  Weill Medical College of Cornell University
Contact Info: http://cbio.mskcc.org/~lianos/contact