2 Classification Task
Assume that we have l training instances xi in an n-dimensional Euclidian space
Rn. Each training instance has a class label yi ∈ Y, where Y = {−1, +1}. The
class labels separate the training instances into two sets I+ and I− (xi ∈ I+
iff yi = +1; xi ∈ I− iff yi = −1). Given a hypothesis space H of all possible
functions h (h : Rn → Y), the classification task is to find a function (classifier) h
that accurately classifies future, unseen data