The structure of a simple back propagation neural networks is shown. We have an input layer (X is the vector of the input i.e. design variables). Then we have a number of hidden layers and an output layer. The output layer gives us the vector Y, that is the vector of the objective functions.

---