Linear Programming task

'Suppose that we have data points (xi,yi) for i ∈ [N], where xi ∈ Rn are feature vectors and yi ∈ {±1} are binary outputs. The linear classification problem aims to find a linear function f(x) = a⊤x + b such that the sign of f(xi) agrees with yi for as many data points i ∈ [N] as possible. You have seen in QBUS1040 the least squares classification technique, which is to get a and b by minimizing �i∈[N](a⊤i x + b − yi)2.
In this question, we will explore two more popular techniques: the support vector machine and the hard margin classifier'