machine learning - Why does support vectors in SVM have alpha (Lagrangian multiplier) greater than zero? -
i understood overall svm algorithm consisting of lagrangian duality , all, not able understand why particularly lagrangian multiplier greater 0 support vectors.
thank you.
i can't figure out too...
if take simple example, of 3 data points, 2 of positive class (yi=1): (1,2) (3,1) , 1 negative (yi=-1): (-1,-1) - , calculate using lagrange multipliers, perfect w (0.25,0.5) , b = -0.25, 1 of our alphas negative (a1 = 6/32, a2 = -1/32, a3 = 5/32).
Comments
Post a Comment