Perceptron algorithm in rules

• weight change = some small constant ? (target activation - spontaneous output activation) ? input activation

• if speak of error instead of the `target activation minus the spontaneous output activation', we have:

• weight change = some small constant ? error ? input activation