A Gaussian kernel SVM is trained on a data set using a very large

$$

$\backslash$gamma $.$ The result is shown below:

gamma.png

Consider the point

$$

$$

$($

$$

$)$

x

$($i$)$

labeled above. If the blue points have label $1,$ and the red points have label $-1,$ what will be the learned weight

$$

$$

$\backslash$alpha

i

$$

corresponding to this data point, approximately?

You may assume that

$$

$\backslash$gamma is as large as you'd like.