The artificial neuron has the main characteristics of the biological neuron:

• the inputs are defined using a vector of inputs noted x,
• the synaptic weights are set to another vector w.

Its values can be positive if they have an activation role or negative if they have an inhibition role.

Its representation is given on the following figure:

The expression of the output is:

According to the biological model, the function f(), called function of activation, is a function of threshold which activates or not the neuron: below the threshold, the neuron is inactive, above, it is activated.

This model is called perceptron. Its simplicity limits it to linearly separable problems. Using a perceptron having two inputs, a line `` '' allows the possible cases to be separated.

In the case of one logical AND or one logical OR, the discrimination of the various combinations is possible, but a function as simple as exclusive OR cannot be solved by a perceptron. See Mr. Minsky and S. Papert (1969) [MP69].