Biological Computation

158    ◾    Biological Computation

Note that even when the network has more than one output neu-

ron, the computation for each output neuron is independent from the

computations for the other output neurons (as each output neuron has

a separate weight vector w), and therefore one can generalize the discus-

sion of a perceptron with a single output neuron to perceptrons with a

higher number of output neurons, such as the perceptron described in

Figure4.4(a).

4.4.3  The Perceptron Learning Rule

We have seen that if the input can be linearly separated a weights vector

achieving the separation is present. In this section we limit our discus-

sion to problems that are linearly separable and present a learning rule

for a perceptron with a sing ...