How to use the generated Neural Fitting Matlab function using "Deploy Solution"...

Hi all, I am a newbie in Matlab so I am sorry if I am asking a silly question. The version I am using is R2014b. I have a matrix "input" (3 variables and 729 samples) and a matrix "target" (3 variables with the 729 values.) I have used the nftool and after several trials let's say I have found something that could be useful. In the "Deploy Solution" I have saved the "Application Deployment" Matlab Function. I thought that I could pass to the generated function the "input" matrix as an argument to generate the "output" and compare it to the "target". Something like:

output = Function(input);

but it does not work. I get the following error message:

" Error using bsxfun Non-singleton dimensions of the two input arrays must match each other.

Error in Function>mapminmax_apply (line 56) y = bsxfun(@minus,x,settings_xoffset);

Error in Function (line 40) xp1 = mapminmax_apply(x1,x1_step1_gain,x1_step1_xoffset,x1_step1_ymin); "

Can anyone help me with that? How can I pass the "input" matrix to the "Function"?

Thanks,

Hector

Answer 1

[Sean de Wolski]

First, don't call it function because that's a keyword. Second, the most common reason I see that is because of a missing transpose:

output = nn_function(xinput')

Answer 2

[Hector Gomez]

Thank you very much Sean. This was exactly the reason why it did not work! Another example of my blindness to certain obvious things together with my ignorance...

Cheers,

Hector

[Sean de Wolski]

Well Neural Nets are the only thing in MATLAB that expect samples to be columns rather than rows. It's confusing and we all do this!