Test trained network with different input sizes

Hello everyone,

My question is if we trained a NN with dataset of size(200 * 50) and targets of size(200 * 50) and the trained procedure is row based (i.e. input1 size = 25 and input2 size = 25 gives the ouput with size = 50).

so after we trained the network and save it to use it as function. can we test it like:

Res = net([input1, input2]); % input1.size = 15 and input2.size = 10
% so the result will have size?

is it possible to test with different size inputs?

Answer 1

[Greg Heath]

All input vectors must have the same length as the training input column vectors. Then the output vectors will always have the same length as the training output column vectors.

Hope this helps.

 *Thank you for formally accepting my answer*

Greg

[amin sari]

I knew that but i want to know if there is an exception!!! i accept it anyway.

thank you dear Greg