Hello to you all,
I'm using MATALB 7 and NN for classification problem which using a large set of data for train.
the data is a set of sampling some process so my data is very much alike; Therefore i can reduce the size of my matrix.
I use the function "prepca" for doing this(after i used prestd) which reduced the size but the matrix i get has values that i can't use (if i train my NN with this matrix and than i will to simulating with vector that has been taken from my data, i will, for sure, get a false detection).

Is there any way to reduce the size and keeping my values, or a way of taking my input vector and change it so it will be valid for the NN?

please help me, if anyone knows,
And thank you in advance.