%%% converts 0,1 class labels to -1,1 %% and calls Lagrangian SVM code by O. L. Mangasarian and D.R. Musicant %% %% input: Xs,Ys, and costs for crossing the margin %% output: weights + bias term wx+b=0 is the separating hyperplane function [w, b] = svml(X,Y_in,cost) %%% convert 0 1 class labels to -1 +1 labels Y=2*Y_in - ones(size(Y_in,1),1); D=diag(Y);; A=X; nu=cost; %% itmax itmax=10000; %% tolerance tol=0.005; %%% call Lagrangian SVM by O. L. Mangasarian and D.R. Musicant [it, opt, w, b] = svml_itsol(A,D,nu,itmax,tol); return;