Toy representation of a linear partition of aCGH samples using large-margin techniques.

The algorithm finds a linear function that is able to partition the aCGH samples into two groups. By solving an optimization problem, the algorithm determines the vector , which geometrically represents the normal vector of a hyperplane (shown in red) separating the samples, along with the bias term , and the assignment of samples to groups. In the toy example shown, the hyperplane separates the samples that present a deletion on the q arm (above the hyperplane) from the ones that do not (below the hyperplane).