As we saw in the previous lecture, solving this optimization recovers a linear classifier of the form y = sign(w ·h(x)+w0) that minimizes the hinge loss for all misclassified points and maximizes the size of the margin (the distance to the closest point to the decision boundary). The term “support vector” refers to the vectors from the decision boundary to the closest points. Note that moving a...

Improperly efficient solutions in the sense of Geoffrion linear fractional vector optimization problems with unbounded constraint sets are studied systematically for first time this paper. We give two conditions which assure that all a given problem improperly efficient. also obtain necessary an solution to be As result, we have new sufficient Geoffrion’s proper efficiency. The obtained results...