Adaline was a competing machine learning algorithm shortly after the

Perceptron was published. It more directly uses gradient descents and has a

linear activation function rather than the ReLU activation function used in the

Perceptron. Note that the weights are updated for all cases not just the error

cases. Like the Perceptron it has a quantizer to translate the linear output into

$1$ and $-1,$ which is explicitly shown in the flow here. Use a least$-$squares cost

function for your algorithm. Study the Adaline flow shown before and try to

understand the differences between it and the Perceptron flow.