I was going for the theory and maths behind the online perceptron algorithm and it is very easy to under stand it intuitively that on a positive mistake, you just add the
x value to the
w and calculate new values for
w and do subtraction in the case of negative mistake
So, in both cases we move closer by 1 to the value we wanted and kinda keep rotating the plane to some degree for infinite until you get a plane that separates the two classes IFF they are linearly separable.
But then I got to know about
Dual Perceptron where learning rate is 1 and you add the counter every time it makes a mistake. How does this algorithm work. Just in simple case intuitively. For maths, I have the original research papers.