comparison toolboxes/distance_learning/mlr/dual/dualWDiagMKL.m @ 0:e9a9cd732c1e tip

first hg version after svn

+function W = dualWDiagMKL(alpha, Z, U, RHO, K)
+
+    global PsiR;
+    d = size(K, 1);
+    m = length(alpha);
+
+    nKernels = size(K,3);
+
+    W = Z - U;
+    for p = 1:nKernels
+        W(:,p) = W(:,p) - diag(K(:,:,p)) / RHO;
+    end
+
+    for i = 1:m
+        W = W + alpha(i) * PsiR{i} / RHO;
+    end
+
+end