Pattern Recognition with Perceptron

Hi, all

I have six patterns as shown below

A1 = [ 0 0 1 1 0 0 0;
       0 0 0 1 0 0 0;
       0 0 0 1 0 0 0;
       0 0 1 0 1 0 0;
       0 0 1 0 1 0 0;
       0 1 1 1 1 1 0;
       0 1 0 0 0 1 0;
       0 1 0 0 0 1 0;
       1 1 1 0 1 1 1];
B1 = [ 1 1 1 1 1 1 1;
       1 0 0 0 0 0 1;
       1 0 0 0 0 0 1;
       1 0 0 0 0 0 1;
       1 1 1 1 1 1 1;
       1 0 0 0 0 0 1;
       1 0 0 0 0 0 1;
       1 0 0 0 0 0 1;
       1 1 1 1 1 1 1];
C1 = [ 0 0 1 1 1 1 1;
       0 1 0 0 0 0 1;
       1 0 0 0 0 0 0;
       1 0 0 0 0 0 0;
       1 0 0 0 0 0 0;
       1 0 0 0 0 0 0;
       1 0 0 0 0 0 0;
       0 1 0 0 0 0 1;
       0 0 1 1 1 1 0];
A2 = [ 0 0 0 1 0 0 0;
       0 0 0 1 0 0 0;
       0 0 0 1 0 0 0;
       0 0 1 0 1 0 0;
       0 0 1 0 1 0 0;
       0 1 0 0 0 1 0;
       0 1 1 1 1 1 0;
       0 1 0 0 0 1 0;
       0 1 0 0 0 1 0];
B2 = [ 1 1 1 1 1 1 0;
       1 0 0 0 0 0 1;
       1 0 0 0 0 0 1;
       1 0 0 0 0 0 1;
       1 1 1 1 1 1 0;
       1 0 0 0 0 0 1;
       1 0 0 0 0 0 1;
       1 0 0 0 0 0 1;
       1 1 1 1 1 1 0];
C2 = [ 0 0 1 1 1 0 0;
       0 1 0 0 0 1 0;
       1 0 0 0 0 0 1;
       1 0 0 0 0 0 0;
       1 0 0 0 0 0 0;
       1 0 0 0 0 0 0;
       1 0 0 0 0 0 1;
       0 1 0 0 0 1 0;
       0 0 1 1 1 0 0];

I have to recognize these patterns with artificial neural network.

I am new in Matlab. Please help!

I need to divide this data into 2 groups.

The first group A1, B1, C1 as training data. The second group A2, B2, C2 used to validate/test the network.

Example : if I select A1 then the output must display 'A', if I select B1 then the output must display 'B', if I select A2 then the output must display 'A'.

. . # # . . .   
. . . # . . .   
. . . # . . .   
. . # . # . .   
. . # . # . .   => This pattern should be recognized as A
. # # # # # .   
. # . . . # .   
. # . . . # .   
# # # . # # #

How do I do that?

Thanks in advance!

Network type is perceptron

Answer 1

[Chandra Kurniawan]

Hi, Benard

A perceptron can be created with the newp function.

Eq: net = newp(PR,S,TF,LF);

- PR is m x 2 matrix of [min max] of network input

- S is a scalar that indicates number of target/pattern

- TF is transfer function. Default is 'hardlim'

- LF is learning function. Default is 'learnp'

In order to divide your data into 2 groups, you need to select A1, B1, C1 as network input.

p = [A1(1:end); B1(1:end); C1(1:end)]';

And to create the network target

t = eye(3);

Then, use train to perform network training. You can also set the network epoch by :

net.trainParam.epochs = 10;

So, the complete code is :

p = [A1(1:end); B1(1:end); C1(1:end)]';
t = eye(3);

PR = zeros(63,2);
PR(:,2) = 1;
net = newp(PR,3,'hardlim','learnp');
net.trainParam.epochs = 10;
net = train(net, p, t);

And about how to test your network just use sim command.

a = sim(net, A2(1:end)');

Now, do testing for A1 and A2. Both of them will display :

>> a = sim(net, A2(1:end)')

a =

       1
       0
       0

>> a = sim(net, A1(1:end)')

a =

       1
       0
       0

[Bernard]

Thank you, sir

But I want the result like 'A','B', or 'C'

Thank you!

[Chandra Kurniawan]

Alist = ['A' 'B' 'C'];

a = sim(net, C1(1:end)');

Alist(find(a))

Is this works?

[Chandra Kurniawan]

You can also change C1 with C2, A1, B2, ...

[Bernard]

Thank you!

[Greg Heath]

If you have the typical problem of identifying the 26 alphabet characters and 10 digits, you will probably need to add a hidden layer via newff or patternnet.

Hope this helps.

Greg