Cellular neural network training

I have two sets of data, the first set is three 3x3 matrices: diagonal and the other two are diagonal-like by means one of the elements in the diagonal is not one.

The second set is three 3x3 matrices of cross and cross-like.

My question is I want the CNN's output to be +1 when I apply any matrices from set1 and -1 for any matrix from set 2,so how can I start to learn the CNN? Shall I apply only the first set and fix the templates or I should apply both sets to the CNN? and if so how can the CNN know during the programming which the desired output is?

Also I want to train the network using Matlab but I didn't use the simulink before, can I just program it? or I must use simulink? what do you advice me? and in case I want to program everything can I use object oriented programming to code and simulate the cells simultaneously?

thank you

[Anh Duong]

Hi,

Have you trained the CNN on the MATLAB yet? I also want to discuss about this. :D

[sali]

Hi

Yes I already trained it in Matlab and Simulink

[Anh Duong]

Currently, I have to trained it for my research, However I havent got a result. Can you help me ? Send me the your code and program? Thanks you so much!

My email is: [e-mail]