Suppose the perceptron algorithm has computed a classifier parameterized by w$.$ Now, consider a new weight vector w$=$ cw where c is a nonzero scalar. Assume that the bias is zero. Using w$$ instead for classification might result in a different decision function