Single-Layer Perceptron Neural Networks
A single-layer perceptron network consists of one or more artificial neurons
in parallel. The neurons may be of the same type we've seen in the
Artificial Neuron Applet.
The perceptron learning rule, which we study next, provides a simple algorithm
for training a perceptron neural network. However, as we will see, single-layer
perceptron networks cannot learn everything: they are not computationally
complete. As mentioned in the introduction, two-input networks cannot approximate
the XOR (or XNOR) functions. Of the (22)n or 16 possible
functions, a two-input perceptron can only perform 14 functions. As the
number of inputs, n, increases, the proportion of functions that can be
computed decreases rapidly.
Each neuron in the layer provides one network output, and is usually connected
to all of the external (or environmental) inputs.
The applet in this tutorial is an example of a single-neuron, single-layer
perceptron network, with just two inputs.
Later, we will investigate multilayer perceptrons.
[Back to the Simple Perceptron
Learning applet page ]