train a neural network in matlab

my project is about facial expression recognition.

for the classification of the emotions i want to train a neural network having an input 20 points taken from the whole face by the vision.CornerDetector function. so each point feature is represented by its X and its Y.

my database contain 110 smiling photos and 95 neutral ones. so that, the whole matrix of smiling faces, for example, has a size of 20x220.

any ideas about how to train my neural network using these features?

[Saud Alramzi]

Hello,

I am working on the same project, did it work for you by chance?

Thanks, Saud

Answer 1

[Greg Heath]

Use patternnet

Ns = 110, Nn = 95, N=Ns+Nn = 205 (NOT 220), I = 20, O = 2

[ I N ] = size(input)

[ O N ] = size(target)

with

target = [ targets targetn ]

targets = repmat([1;0],1,Ns)

targetn = repmat([0;1],1,Nn)

Hope this helps.

Thank you for formally accepting my answer

Greg

[rana]

thank you :)