These exercises provide us with additional experience using the multilayer perceptron. In this case, we will be training a network to identify patterns that are symmetric about their centre. In one exercise, we will be interested in exploring the structure of a network that can do this. In a second exercise, we will push the limits of this sort of training, by using a single hidden unit. This will give some valuable hands-on experience with respect to recognizing situations in which learning is stalling.