wrong values from output of neural network?

i train a neural network for classification but when i tested it i got wrong values

load ('net.mat');
p=sim(net,testfeatures)

the answer:

p =

      0.1422
      0.1371
      0.7207

p must have one one'1' and two zeros'0' or what this numbers mean?

[Walter Roberson]

That depends on how you constructed the net, which is information you did not include.

Notice that round(p) gets you the information you are expecting.

[samar mahmoud]

thanks can you explain in general what is the output of trained neural network thanks in advance

[Walter Roberson]

For two-class networks it is frequently a linear scale representing the probability that it belongs to the second class (so less than 1/2 represents that it is more likely to be long to the first class.)

But it depends how it was built. In other cases you might get a list of values, one per class, representing the relative likelihood that the item belongs to that class.