To my understanding, you have trained a network using patternnet but while testing the simulated response, you are getting only one answer.
Please check the size of features and see if it contains only 1 example. Also, while training, you have used T’, which is transpose of the matrix T, but while checking performance, you are comparing T and y. Usually, we pass the same matrix in both the places of size NumberOfClasses x NumberOfExamples. PFA sample code:
[x,t] = iris_dataset;
net = patternnet(10);
net = train(net,x,t);
view(net)
y = net(x);
perf = perform(net,t,y);
classes = vec2ind(y);
out = sim(net, x);
