def fit$($self$,$ X$,$ y$,$ pocket $=$ True, epochs $=100)$:

$\text{'}\text{'}\text{'}$ find the classifer weight vector and save it in self.w

X: n x d matrix, i$.$e$.,$ the bias feature is not included.

It is assumed that X is already normalized be data preprocessing.

y: n x $1$ vector of $\{+1,-1\}$

degree: the degree of the Z space

return self.w

$\text{'}\text{'}\text{'}$