Overview
Theorem
Given a loss function {% \mathcal L %} and a regularizer {% R %}, the minimizer of the following
{% \mathcal{L}(\vec{x}_i,y_i,f(\vec{x}_i)) + R(||f||) %}
has a representation as
{% f = \sum_i \alpha_i k(\vec{x}_i,) = \sum \alpha_i \phi(\vec{x}_i) %}
(see Ye)