Adaline model
The Adaline model is a form of an artificial neural network . The name stood for ADAptive LInear NEuron , when neural networks later lost their popularity, the acronym was also translated as ADAptive LINear Element .
The Adaline model was proposed in 1959 by Bernard Widrow and Marcian Edward Hoff and forms a perceptron in which the learning algorithm can also activate an output unit as information. The picture shows the structure of an ADALINE. The learning algorithm adapts the weights of the input vector and the bias (tendency) input.
As a learning rule, Widrow and Hoff proposed the so-called delta rule (also known as the Widrow-Hoff rule or the least mean squares algorithm ). This causes the artificial neural network to behave differently in that it builds up more stable knowledge, as the modified learning rule means that knowledge that has been learned is less likely to be forgotten again.