ABSTRACT
Now we turn to the linear regression problems. The linear support vector regression is established by converting the linear regression problems to the linear classification problems.
Similar to classification problems, regression problems consist of finding a real function, for a given training set T: T = {(x1, y1), · · · , (xl, yl)}, where xi ∈ Rn is an input, and yi ∈ Y = R is an output, i = 1, · · · , l. Rather than just Y = {−1, 1} in classification problems, Y is generalized to the real set in regression problems. Correspondingly, the goal of regression problems is to derive the real value of an output y for any input x, based on a training set T.
