#margin

Question 1: Single choice

Training data: negatives $(-1, 0)$ and $(2, -2)$ , positive $(1, 0)$ (labels $-1$ and $+1$ ). A linear SVM is written as $h(\mathbf{x}) = \mathbf{w} \cdot \mathbf{x} + b$ in canonical form: $y_i\,h(\mathbf{x}_i) \ge 1$ for all points, with equality for the support vectors, and $\lVert \mathbf{w} \rVert$ as small as possible. Which $(\mathbf{w}, b)$ is the SVM solution?

$\mathbf{w} = (1, 2)$ , $b = 0$ 1
$\mathbf{w} = (1, 1.5)$ , $b = 0$ 2
$\mathbf{w} = (1, 1)$ , $b = 0$ 3
$\mathbf{w} = (2, 3)$ , $b = 0$ 4

Timed test