To minimise the loss function in support vector machine learning each data point is associated with a Lagrange multiplier an. At the end of the training some of the an=0, others not. A. The final learned weight vector depends on all of the data points. B. If an=0 the corresponding data point is on the margin hyperplane. OC. The weight vector depends only on the data points for which an +0. OD. If Q, 70 the corresponding data point is a support vector. E. If an=0 the corresponding data point xn is correctly classified and away from the margin hyperplanes. OF. There are multiple weight vectors that solve the SVM learning task