I made a patternnet, and i cannot understand what target values mean

I have a set of image feature data and i have made my target matrix as:

targets = zeros(3,900);

targets(1,1:300) = 1;

targets(2,301:600) = 1;

targets(3,601:900) = 1;

what would change if i changed the 1 values as:

targets = zeros(3,900);

targets(1,1:300) = 1;

targets(2,301:600) = 2;

targets(3,601:900) = 1;

can anyone explain me what is their role?

Answer 1

[Anshika Chaurasia]

Hi Sougles,

In the following code, the dataset labels/target/classes are in one-hot encoding format.

targets = zeros(3,900);
targets(1,1:300) = 1;
targets(2,301:600) = 1;
targets(3,601:900) = 1;

Looking at the shape of targets, there are 3 classes in your dataset.

Let's say there are three classses A, B and C. So, we will need three binary variables for one-hot encoding. A "1" value is place in the binary variable for the class and "0" for the other classes.

Hence,

targets(1,1:300) = 1; % samples 1 to 300 belong to first class
targets(2,301:600) = 1;% samples 301 to 600 belong to second class
targets(3,601:900) = 1;% samples 601 to 900 belong to third class